<< 24-04-2018 >>

---

00:00:34	FromGitter	<zacharycarter> like I mean almost every web app / enterprise needs a database
00:00:48	FromGitter	<zacharycarter> and the DB support we have is built into the stdlib
00:00:52	FromGitter	<zacharycarter> and then there's ormin
00:01:16	FromGitter	<zacharycarter> but I mean - if a big company is evaluating Nim and that's the level of support for database access they have - they're not going to use Nim
00:02:45	FromGitter	<zacharycarter> it's also a chicken / egg scenario - because if no one is really using Nim for these things - how does the support for them arise in the first place>
00:09:34	*	simonsolnes quit (Ping timeout: 260 seconds)
00:10:18	*	jrbrt quit (Quit: jrbrt)
00:11:38	*	jrbrt joined #nim
00:11:42	FromGitter	<krux02> nim is not for big companies
00:11:59	shashlick	bodie_: skrylar spent some time on it https://github.com/Skrylar/skflatbuffers
00:12:22	FromGitter	<krux02> it cant provide the support
00:12:23	FromGitter	<zacharycarter> it could be though
00:12:27	FromGitter	<zacharycarter> it doesn't need to
00:12:38	FromGitter	<honewatson> yeah it could be
00:12:56	FromGitter	<zacharycarter> I mean - we're thinking about leveraging erlang-OTP and elixir at work
00:13:09	FromGitter	<zacharycarter> we don't expect any kind of support through that making that decision
00:13:24	FromGitter	<zacharycarter> you use OS software at your own risk - the choice boils down to what suits your needs the best
00:13:46	FromGitter	<zacharycarter> Nim is well maintained and has an active SDLC and fits certain problem domains exceptionally well
00:13:55	FromGitter	<zacharycarter> and support grows with the user base
00:14:22	skrylar	i forget what is missing from that lib
00:14:28	FromGitter	<honewatson> Nginx really became big after version 1. Got turned into a business etc I think I started using it around 0.5 and it was interesting watching the transformation
00:14:36	FromGitter	<zacharycarter> you could make exactly the same assertion with Rust if they didn't have Mozilla as an original backer
00:14:40	skrylar	it does work, although i think its missing flexbuffers and automatic serializatoin
00:14:58	skrylar	have been writing a protobuf compiler in smalltalk as of late, admittedly
00:15:17	FromGitter	<zacharycarter> smalltack is tits
00:15:22	FromGitter	<zacharycarter> smalltalk*
00:15:28	FromGitter	<zacharycarter> but I think you need to come back to Nim
00:15:30	FromGitter	<zacharycarter> :P
00:15:47	skrylar	going back outside the classbrowser sucks
00:16:02	FromGitter	<krux02> nim won't be used by big companies, because there are just not enough nim developers and nim training programs around.
00:16:04	FromGitter	<zacharycarter> :/
00:16:06	skrylar	have to do it to work with godot and i'm constantly cringing at writing utility methods instead of bolting them on to where they belong
00:16:13	FromGitter	<zacharycarter> krux02: that's also a misconception
00:16:24	FromGitter	<zacharycarter> you don't need every developer at your org to know every language
00:16:47	FromGitter	<zacharycarter> you can hire a bunch of mediocre java / .NET / whatever devs to build a certain portion of your stack
00:16:54	FromGitter	<zacharycarter> but you can't build your entire stack with them
00:16:58	FromGitter	<krux02> big companies won't suddenly choose to developm in Nim.
00:17:00	FromGitter	<zacharycarter> otherwise you end up with a mediocre stack
00:17:14	FromGitter	<honewatson> I think Nim can work well in the start up space
00:17:16	FromGitter	<zacharycarter> well I've tempted carfax by saying - if Nim eventually gets AWS support
00:17:23	FromGitter	<zacharycarter> and a few other techs we rely on
00:17:27	skrylar	what do you need for AWS support; some serialization thing?
00:17:34	shashlick	the foot in the door will happen when it does one thing super well - one use case where it is super ready
00:17:35	FromGitter	<zacharycarter> it'd be a better alternative to Rust / C / C++ / etc...
00:17:38	FromGitter	<krux02> it will be small indie projecs or some experts who are frustrated with the established tools that will come to the Nim community.
00:17:47	FromGitter	<krux02> And they will mostly be alone
00:17:58	FromGitter	<zacharycarter> skrylar: Varriount and I are working on marshalling / unmarshalling and codgen from the botocore stuff
00:18:16	FromGitter	<krux02> Nim just needs to support such individuals to compete agains projects written with bigger teams
00:18:18	skrylar	hum. someone should port petitparser2 :)
00:18:45	FromGitter	<zacharycarter> never heard of it but I will look into it :)
00:18:53	skrylar	its a packrat/peg parser
00:19:09	FromGitter	<zacharycarter> I just don't think we should proceed with the conception that Nim can't become a language that enterprise or large orgs consider
00:19:13	skrylar	been using it in Pharo (smalltalk) to write the protobuf compiler
00:19:27	FromGitter	<zacharycarter> that's a self-defeating viewpoint IMO
00:19:27	FromGitter	<krux02> @zacharycarter did you see I have recursive ast matching in my ast_pattern_matching library
00:19:31	FromGitter	<zacharycarter> we should figure out how we can get there
00:19:33	CodeVantage_	Here's a question, in which area does nim give you a competitive advantage?
00:19:38	skrylar	trouble with adoption: you need a GUI that isn't crud (nim fails here) and/or the ability to interact with existing service (i dunno. pharo has this trouble too)
00:19:55	FromGitter	<zacharycarter> @krux02 - I did - I haven't gotten to play around with it yet but I'm super excited to
00:19:58	FromGitter	<krux02> I use emacs, that is similar to Pharo/smalltalk
00:20:04	FromGitter	<zacharycarter> definitely when I start rewriting frag I will leverage it
00:20:06	skrylar	i did do that fltk wrapper that nobody uses, and a firebird that nobody uses, and got a little tired of spending some months effectively doing nothing (or feels like it)
00:20:13	FromGitter	<krux02> Pharo is a smalltalk vm and everything is smalltalk, emacs is a lisp-machine
00:20:20	FromGitter	<zacharycarter> anyway - afk for a few
00:20:51	skrylar	we need an approved gtk3 ffs. the stdlib is on 2, worlds moved on from that
00:20:56	FromGitter	<zacharycarter> skrylar: you sould work with @Varriount and I
00:21:12	FromGitter	<zacharycarter> I think you have more need for certain techs than we do
00:21:28	FromGitter	<zacharycarter> but I don't think that's any reason preventing us from collaborating on some more libs for Nim
00:21:52	FromGitter	<krux02> I will only develop what will also benefit me personally, and I don't like gui toolkits.
00:21:55	skrylar	the firebird one was for a request here
00:22:12	FromGitter	<krux02> I am much better off implementing button, on my own.
00:22:15	skrylar	wrapped most of the C lib, also had to write and scavenge docs for the whole thing
00:22:25	FromGitter	<zacharycarter> if I could lead such a life @krux02 ... :P
00:22:34	FromGitter	<zacharycarter> I can't employ myself that way though yet unfortunately
00:22:53	FromGitter	<krux02> well, to be honest I am pretty poor at the moment
00:22:58	FromGitter	<zacharycarter> hrm - well again I think we have a lot of people using nim in very niche use cases
00:23:11	FromGitter	<zacharycarter> we need to figure out like - what does everyone want from Nim potentially - and then build that
00:23:12	skrylar	interesting amount of game developers were interested in nim for a while there
00:23:21	FromGitter	<zacharycarter> <- one of them
00:23:23	FromGitter	<zacharycarter> also @krux02
00:23:32	FromGitter	<krux02> But well, yea, we'll see how much I can do for the Nim community, when I actually get paid to not program in Nim.
00:23:51	FromGitter	<zacharycarter> yeah - it's like program in this shitty language to make money
00:24:15	FromGitter	<krux02> exactly
00:24:26	FromGitter	<zacharycarter> I was explaining to my manager though
00:24:29	FromGitter	<krux02> probably javascript, even though I will avoid that if I can
00:24:31	FromGitter	<zacharycarter> that's how you build a shitty engineering org
00:24:42	FromGitter	<krux02> but to be honest, no
00:24:51	FromGitter	<krux02> I think I can afford saying away from that
00:24:54	FromGitter	<zacharycarter> you'd be better off hiring C++ programmers fresh out of school
00:25:08	FromGitter	<krux02> there are just too many people around doing web fuzz
00:25:10	FromGitter	<zacharycarter> than hiring a java developer that doesn't understand what a pointer is
00:25:12	FromGitter	<zacharycarter> or a linked list
00:25:14	FromGitter	<zacharycarter> yeah
00:25:16	FromGitter	<krux02> I don't need to dig into that
00:25:28	FromGitter	<zacharycarter> but the thing is there too - you grow from there by adding complexity
00:25:29	FromGitter	<krux02> let somebody else fuck up the frontend code
00:25:34	FromGitter	<zacharycarter> like maybe today you're writing web apps
00:25:40	FromGitter	<zacharycarter> but then you start gathering data
00:25:49	FromGitter	<zacharycarter> so you need to start leveraging ML
00:25:55	FromGitter	<zacharycarter> and then you want to optimize your data pipeline
00:25:58	FromGitter	<krux02> I honestly don't see myself writing web apps.
00:26:05	FromGitter	<zacharycarter> I'm not sayhing you
00:26:10	FromGitter	<zacharycarter> I'm just saying enterprises in general
00:26:20	FromGitter	<zacharycarter> that's how AWS goes from being an ecommerce book selling web site
00:26:24	FromGitter	<zacharycarter> to the pinnacle of enterprise IT
00:26:32	FromGitter	<zacharycarter> they didn't stay in that ecommerce webdev model
00:26:46	FromGitter	<zacharycarter> they added complexity and yearned for results through software dev
00:26:50	CodeVantage_	i thought amazon used a lot of money
00:26:54	*	jrbrt quit (Quit: jrbrt)
00:26:56	FromGitter	<krux02> The problem is that, by that point in time the codebase is su much screwed up and horrible to work with that nobody want to do it.
00:27:00	FromGitter	<zacharycarter> well I mean jeff bezos owns the washington post
00:27:06	FromGitter	<krux02> And some asshole then needs to dig through the shit.
00:27:18	FromGitter	<zacharycarter> well it's more about the asshole saying that it needs to be the way it is
00:27:33	FromGitter	<zacharycarter> like we have people at my company saying every single piece of data we store needs to be encrypted
00:27:39	FromGitter	<zacharycarter> wtf kind of crack are you smoking?
00:27:48	CodeVantage_	lol
00:27:55	FromGitter	<zacharycarter> especially when you're moving to cloud based infrastructure where encryption costs money
00:28:17	CodeVantage_	either its useless encryption ( too much work and not secure enough)
00:28:22	FromGitter	<krux02> well, encryption is actually important.
00:28:22	CodeVantage_	or you can't do anything
00:28:30	FromGitter	<zacharycarter> you don't encrypt data you can easily google
00:28:34	FromGitter	<zacharycarter> is my point
00:28:42	FromGitter	<zacharycarter> which is what ppl at my company are suggesting we do
00:28:49	FromGitter	<krux02> but the way cloud companies work today, encryption is most of the time just an illusion
00:28:59	CodeVantage_	Well maybe they meant the whole compute should be encrypted?
00:29:05	FromGitter	<krux02> an illusion to make the people feel safe
00:29:05	FromGitter	<zacharycarter> nah
00:29:06	FromGitter	<zacharycarter> it's like
00:29:17	FromGitter	<zacharycarter> we can't use dynamodb because it doesn't support encryption at rest
00:29:33	FromGitter	<zacharycarter> guess what - I'm pretty sure there are lots of companies w/ more sensitive data than carfax, leveraging dynamodb
00:29:59	FromGitter	<zacharycarter> it's just stupid over engineered standards that add complexity to our solutions
00:30:10	CodeVantage_	yes. they use security guards
00:30:10	FromGitter	<zacharycarter> we didn't go to the could by starting out with like digital ocean or heroku or something
00:30:13	FromGitter	<krux02> copanies keep a backdoor in case the customer looses their key. In other words they have the key to the encrypted data next to the encrypted data. And that basically counts as not encrypted
00:30:22	CodeVantage_	like physical people
00:30:27	FromGitter	<zacharycarter> it's like - let's go to the cloud, use every AWS service, slap kubernetes on top of it, manage everything through terraform etc
00:30:55	FromGitter	<zacharycarter> and then we'll take tools specifically designed for a task, and then manage them through another tool - not designed for that task
00:30:55	FromGitter	<krux02> what has terraforming to do with web development?
00:31:13	FromGitter	<zacharycarter> so we managed kubernetes on aws through kops- but then have to manage kops through terraform
00:31:26	FromGitter	<zacharycarter> it literally makes me want to jump out of a window most days
00:31:30	FromGitter	<zacharycarter> but I'm afraid I won't die on impact
00:31:37	FromGitter	<krux02> I don't know those tools, and I kind of don't want to.
00:31:46	FromGitter	<zacharycarter> I suggest you stay ignorant
00:31:58	FromGitter	<zacharycarter> kubernetes is nice if you use docker
00:32:02	FromGitter	<zacharycarter> otherwise just forget it existed
00:32:06	CodeVantage_	hmmm . is the person in charge not that technically minded?
00:32:16	FromGitter	<krux02> yea, I take the second one.
00:32:21	FromGitter	<zacharycarter> well our architects often purport having worked w/ a certain tech for years
00:32:25	FromGitter	<zacharycarter> but you google them and find their github
00:32:34	FromGitter	<zacharycarter> and they of course have no repos demonstrating that knowledge
00:32:38	FromGitter	<krux02> docker is something I heared that it exists. I read about it, and then forgot what it was.
00:32:39	CodeVantage_	and?
00:32:48	FromGitter	<zacharycarter> I mean - one of our architects suggested the other day that in order to "automate" something
00:32:52	FromGitter	<zacharycarter> we'd have to run a shell script twice
00:33:10	*	NimBot joined #nim
00:33:23	FromGitter	<zacharycarter> docker is pretty sweet - but if you're looking for a containerization solution - other open source alternatives exist
00:33:43	FromGitter	<krux02> tar.gz?
00:33:43	FromGitter	<zacharycarter> docker really serves the enterprise world - I just use it w/ Nim because I know it
00:34:02	FromGitter	<zacharycarter> no - more like LXC
00:34:16	FromGitter	<krux02> I don't know lxc
00:34:22	FromGitter	<zacharycarter> goolge/ wiki it
00:34:29	FromGitter	<zacharycarter> you might actually find it useful in some capacity maybe
00:34:36	FromGitter	<zacharycarter> or the concept anyway
00:34:50	FromGitter	<zacharycarter> just containerization really
00:35:02	FromGitter	<zacharycarter> evolved / better form of sandboxing
00:35:10	FromGitter	<zacharycarter> everything is ephemeral
00:35:33	FromGitter	<krux02> so it's basically multiple users on the same machine.
00:35:33	FromGitter	<zacharycarter> it's basically how the playground (which is almost done - tonight I hope) works
00:36:02	FromGitter	<zacharycarter> mmm - it's like spinning up an isolated environment for whatever you need
00:36:08	FromGitter	<zacharycarter> that is intended at the end of the day - to be destroyed
00:36:10	FromGitter	<krux02> where useres are "virtual-servers"
00:36:18	FromGitter	<zacharycarter> yeah essentially
00:36:29	FromGitter	<zacharycarter> but I mean - you have a docker daemon / service that runs
00:36:43	FromGitter	<zacharycarter> and manages all these containers
00:36:51	FromGitter	<zacharycarter> and you declare a container via a dockerfile
00:37:02	FromGitter	<zacharycarter> and then you can orchestrate containers via docker-compose
00:37:08	FromGitter	<krux02> I mean "isolated environments" that is basically what users on a linux machine get. Since I know linux.
00:37:15	FromGitter	<zacharycarter> well that's sandboxing
00:37:26	FromGitter	<zacharycarter> because the environment is meant to be persistent
00:37:38	FromGitter	<krux02> the only problem is that users can install software, and that is for servers not really cool.
00:37:39	FromGitter	<zacharycarter> containers are meant to be created and dertroyed
00:37:44	FromGitter	<zacharycarter> well docker solves that
00:37:49	FromGitter	<zacharycarter> or container sdo
00:37:58	FromGitter	<zacharycarter> like - if I want to install mysql on my machine and I don't really want to install it
00:38:01	FromGitter	<zacharycarter> I can use docker for that
00:38:04	FromGitter	<krux02> but I like for that think nixos.
00:38:13	FromGitter	<zacharycarter> yeah nixos is sweet
00:38:31	FromGitter	<zacharycarter> but containers are deployable to various infrastructure
00:38:50	FromGitter	<krux02> honestly, who cares about that?
00:38:52	FromGitter	<zacharycarter> so I might have a docker-compose file that says I have a postgresql database, a nginx reverse proxy, a nim web appp etc
00:38:55	FromGitter	<zacharycarter> well
00:39:02	FromGitter	<zacharycarter> I can take this docker setup
00:39:03	FromGitter	<krux02> you need a stable system that you can deploy to
00:39:13	FromGitter	<zacharycarter> and as long as whatever provider I wants supports docker
00:39:16	FromGitter	<zacharycarter> I can deploy it there
00:39:26	FromGitter	<zacharycarter> that's the thing - you define all that in your docker solution
00:39:40	FromGitter	<zacharycarter> so like - with the playground - I'm saying I have a reverse proxy serving up this porty
00:39:42	FromGitter	<zacharycarter> port*
00:39:46	FromGitter	<zacharycarter> and I have a webapp listening here
00:39:55	FromGitter	<zacharycarter> and I have a database listening on this port etc
00:40:01	FromGitter	<zacharycarter> and then if anything goes wrong with my environment
00:40:05	FromGitter	<krux02> I am not saynig docker isn't useful, but for me that is really something the OS should do.
00:40:07	FromGitter	<zacharycarter> I can just destroy the docker container instance
00:40:13	FromGitter	<zacharycarter> and spin it back up
00:40:25	FromGitter	<zacharycarter> meh - you can't always rely on the fact that your environment is going to match your needs
00:40:35	FromGitter	<zacharycarter> w/ docker - it doesn't matter if the host is windows or osx or linux or whatever
00:40:37	FromGitter	<krux02> then yeas switching the platform and the os would cost time/money. But that is something you wouldn't do too much anyway.
00:40:46	FromGitter	<zacharycarter> in the enterprise world you'd be surprised
00:41:07	FromGitter	<zacharycarter> especially because certain cloud providers - AWS, MS (Azure), Google (GCP) specialize in different things
00:41:37	FromGitter	<zacharycarter> so like - it's a best tool for the job situation
00:41:53	FromGitter	<zacharycarter> and if you can deploy a solution agnostic ally - that's your best bet
00:41:55	shashlick	zacharycarter is it really true that the OS doesn't matter
00:42:03	FromGitter	<krux02> well yea, it's useful. It has it right to exist
00:42:15	FromGitter	<zacharycarter> well as long as it's running docker it doesn't right?
00:42:27	FromGitter	<zacharycarter> because the docker image itself specifies what OS the container runs on
00:42:38	FromGitter	<zacharycarter> I mean - docker doesn't have like windows or osx images tmk
00:42:43	FromGitter	<zacharycarter> I think maybe it has windows images these days
00:42:49	FromGitter	<zacharycarter> but in terms of the host operating system
00:42:53	shashlick	Exactly so the container is still OS specific
00:42:55	FromGitter	<zacharycarter> I don't think it matters much
00:42:58	FromGitter	<zacharycarter> yes the container is
00:42:59	FromGitter	<zacharycarter> but not the host
00:43:14	FromGitter	<zacharycarter> that's what I was trying to get at
00:43:28	FromGitter	<zacharycarter> anwyay - I really need to pee / run downstairs so I'll be back in a bit
00:43:28	FromGitter	<krux02> but to me it sounds a bit like Java, it can be executed on every operating system. But it eats memory and is slow.
00:43:33	FromGitter	<zacharycarter> fuck java
00:43:37	FromGitter	<zacharycarter> and no docker isn't necessarily sow
00:43:43	FromGitter	<zacharycarter> you can run C programs w/ docker
00:43:45	FromGitter	<zacharycarter> or whatever
00:43:46	shashlick	So it's just that they have standardized an image format
00:43:52	FromGitter	<zacharycarter> exacxtly
00:43:55	FromGitter	<zacharycarter> much like LXC
00:44:11	FromGitter	<zacharycarter> docker doesn't do backwards compatibility very well though, and of course not open source - so
00:44:21	*	CodeVantage_ quit (Read error: Connection reset by peer)
00:44:23	shashlick	As an Enterprise you still have to pick that base OS
00:44:27	FromGitter	<zacharycarter> but IMO pick your battles - docker brings a lot to the table that other containerization solutions don't
00:44:36	FromGitter	<zacharycarter> yeah - I mean I use alpine for everything I can
00:44:42	FromGitter	<zacharycarter> because it's so slim
00:44:52	FromGitter	<zacharycarter> the goal of docker isn't to have a robust OS
00:44:58	shashlick	My understanding is that docker solves the distribution challenge
00:45:00	FromGitter	<zacharycarter> it's to build the slimmest runtime that can be shipped for your app
00:46:15	skrylar	rocket was neat
00:46:40	skrylar	is that still around? they were trying to codify something docker-esque as a standard so other people could implement it; there were some imps for linux and also one on freebsd through the jail mechanism
00:47:16	skrylar	always saw docker style stuff a means to deal with things like the java runtime hell more than anything
00:47:38	skrylar	for instance if install java 10, freeplane crashes all the time. while something else may need 10. but two docker images = problem solved
00:48:42	shashlick	Yep, this problem has been solved so many times
00:49:44	shashlick	Anaconda, choosenim, portable apps, etc etc
00:49:52	skrylar	am a fan of appimages
00:50:14	skrylar	Really wish elementary had backed appimagse instead of yet-another-appstore tbh
00:52:24	shashlick	Appimage looks interesting
00:52:43	skrylar	the theory is you just make an executable squashfs file and you run it.
00:52:48	shashlick	Every Linux distribution comes to mind, package manager
00:53:06	skrylar	i think it just does a fuse mount and runs a bash script
00:53:11	shashlick	It just feels like the same problem solved over and over
00:53:11	*	CodeVantage_ joined #nim
00:53:25	shashlick	Nothing really novel has come out
00:53:31	shashlick	Guess I'm getting old
00:53:41	skrylar	eh. diaspora happens
00:54:14	skrylar	when you don't have a powerful leader (Hobbs logic here) then people don't really cohease. and a lot of *nix people run away from attempts at cohesion
00:55:49	*	tersec joined #nim
00:58:31	tersec	I posted a response to https://forum.nim-lang.org/t/3772 a day or so ago, and while I can see it when I'm logged in, I can't when I'm logged out. What's going on?
00:59:05	tersec	Incidentally, https://forum.nim-lang.org/t/3772/3 gets linked to but 404s.
01:01:44	shashlick	You need to have dom96 enable your account
01:04:09	*	CodeVantage_ quit (Read error: Connection reset by peer)
01:08:52	*	CodeVantage_ joined #nim
01:09:04	*	cspar joined #nim
01:13:40	skrylar	i don't see anything special in this botocore yet
01:14:21	skrylar	zacharycarter is this just a bunch of faceplates to json calls?
01:14:40	CodeVantage_	what happened to random?
02:00:48	*	SenasOzys__ joined #nim
02:00:52	*	SenasOzys quit (Remote host closed the connection)
02:16:56	*	dddddd quit (Remote host closed the connection)
02:17:38	*	tiorock joined #nim
02:17:39	*	tiorock quit (Changing host)
02:17:39	*	tiorock joined #nim
02:17:39	*	rockcavera quit (Killed (orwell.freenode.net (Nickname regained by services)))
02:17:39	*	tiorock is now known as rockcavera
02:35:53	*	btbytes joined #nim
02:41:19	*	btbytes quit (Quit: Textual IRC Client: www.textualapp.com)
02:53:38	FromGitter	<zacharycarter> skrylar: yeah - some of the s3 services use XML and others use JSON for their APIs, I think botocore mainly contains the service specifications and utility code
02:53:55	FromGitter	<zacharycarter> boto3 is the python AWS sdk I think - and it's not at all idiomatic python from my understanding
02:54:02	FromGitter	<zacharycarter> more like Java written in Python
02:56:42	skrylar	i don't really need anything to do with AWS, although i guess i could fiddle with it for rep
02:56:56	skrylar	tbh if its just a faceplate, would write spec files and use converter scripts to generate the nim
02:57:25	FromGitter	<zacharycarter> well - I met with Varriount yesterday at starbucks and he identified 3 requirements
02:57:43	*	Snircle joined #nim
02:58:15	FromGitter	<zacharycarter> being able to marshal to specs like - https://raw.githubusercontent.com/boto/botocore/develop/botocore/data/s3/2006-03-01/service-2.json
02:58:22	FromGitter	<zacharycarter> and being able to unmarshal from them
02:58:38	FromGitter	<zacharycarter> and then being able to actually generate the datatypes and operations that they specify
02:58:52	FromGitter	<zacharycarter> I'm not sure what else is involved
02:58:57	FromGitter	<zacharycarter> if anything
02:59:25	skrylar	i seem to somehow keep getting roped in to writing low level bitcoding crap. lol
02:59:43	skrylar	OH i remember
02:59:57	skrylar	i had issues with reflecting data types in flatbuffer
03:02:36	*	Snircle quit (Client Quit)
03:02:52	skrylar	it turns out the protobuf wire format is pretty simple too
03:03:18	skrylar	has there been any progress on being able to shove tags on to object fields and reflect those at compile time (without having to eat the entire typespec in a macro)
03:07:18	FromGitter	<zacharycarter> umm
03:07:22	FromGitter	<zacharycarter> varriount and I were talking about this earlier
03:07:27	FromGitter	<zacharycarter> I don't knwo what tags are
03:07:34	FromGitter	<zacharycarter> we were looking at - https://nim-lang.org/docs/manual.html#implementation-specific-pragmas-custom-annotations
03:07:37	FromGitter	<zacharycarter> is that the same thing?
03:18:48	skrylar	in Go you can use tildes on fields `likeThis` and its called a "tag" which does nothing but put some stuff on a type, which reflection can inspect
03:19:09	skrylar	custom annotations looks like araq got around to doing a good
03:19:18	skrylar	maybe i should go update the flatbuffers
03:19:39	skrylar	could also do native protobuf integration with those
03:23:32	FromGitter	<zacharycarter> :D
03:24:22	FromGitter	<zacharycarter> I am so pissed off right now - I ordered like $250 worth of shirts and hoodies last night off some website and then I opened the e-mail today and realized that I paid $6-$12 for shipping on every gd item
03:24:23	shashlick	can you do a chmod in nim?
03:24:39	FromGitter	<zacharycarter> not sure - I know you can copy with permissions
03:24:44	FromGitter	<zacharycarter> I'm guessing there's a way to
03:25:25	FromGitter	<zacharycarter> shashlick: https://nim-lang.org/docs/os.html#setFilePermissions,string,set[FilePermission]
03:25:29	FromGitter	<zacharycarter> mebbe?
03:25:41	shashlick	yep looks like, trying to figure out the syntax
03:25:46	shashlick	i want 755
03:26:01	FromGitter	<zacharycarter> not sure how it abstracts it across OS's
03:26:14	FromGitter	<zacharycarter> oh
03:26:18	FromGitter	<zacharycarter> I see yeah
03:26:23	FromGitter	<zacharycarter> equating 755 to - https://nim-lang.org/docs/os.html#FilePermission
03:26:24	FromGitter	<zacharycarter> not sure haha
03:26:56	shashlick	:D
03:27:23	FromGitter	<zacharycarter> I don't even know what 755 is - I'm always just like - chmod 777
03:28:37	FromGitter	<zacharycarter> ughhh this website is so deceiving - they have a shopping cart where you add items, but then when you place an order, every item your cart becomes an individual order
03:28:51	FromGitter	<zacharycarter> unless you order more than one of the same thing
03:29:03	FromGitter	<zacharycarter> none of this is explained up front though
03:31:58	*	endragor joined #nim
03:32:06	*	Snircle joined #nim
03:32:53	FromGitter	<zacharycarter> I'm sure my threat of a report to the BBB will scare them shitless and I will win free shipping on my entire order- but only time will tell :P
03:33:23	*	endragor quit (Remote host closed the connection)
03:33:51	*	endragor joined #nim
03:38:03	FromGitter	<zacharycarter> hrm - struggling w/ one last thing with the playground and docker
03:39:08	*	Snircle quit (Quit: Textual IRC Client: www.textualapp.com)
03:42:17	FromGitter	<zacharycarter> nm think I figured it out
03:49:05	shashlick	zacharycarter: let me know if you need any help with that
03:49:57	*	rockcavera quit (Ping timeout: 240 seconds)
03:50:55	FromGitter	<zacharycarter> I think it was just an issue with how I was calling execProcess and the differences between alpine and the container OS I was using previously
03:51:11	FromGitter	<zacharycarter> but if I can't get things to work I'll let you know, and thanks for offering help :D
03:51:31	FromGitter	<zacharycarter> gonna go to sleep, and hopefully wake up early to get some work finished on the playground
03:51:50	FromGitter	<zacharycarter> I'd like to hand it off tomorrow - and then start on revamping the front end to be like the wasm studio
03:51:53	FromGitter	<zacharycarter> but baby steps
03:51:58	FromGitter	<zacharycarter> anyway - goodnight!
03:53:32	*	rockcavera joined #nim
03:53:32	*	rockcavera quit (Changing host)
03:53:32	*	rockcavera joined #nim
03:53:50	CodeVantage_	has anyone ever tried swig to make nim bindings?
03:54:23	CodeVantage_	nite
03:59:18	*	leorize quit (Ping timeout: 255 seconds)
04:01:00	FromGitter	<gogolxdong> How to iterate files under directory?
04:04:40	*	CodeVantage_ quit (Read error: Connection reset by peer)
04:21:40	*	skrylar quit (Quit: Leaving)
04:22:19	FromGitter	<gogolxdong> Do we have to use listFiles in nimscript?
04:24:31	FromGitter	<gogolxdong> or oswalkdir?
04:30:24	FromDiscord	<bodie> are there any decent attempts using Nim to implement a compiler or scripting language?
04:32:09	shashlick	listFiles() in nimscript
04:32:12	shashlick	os won't work
04:32:21	shashlick	bodie_: nim is written in nim
04:32:49	FromGitter	<mratsim> @bodie: https://github.com/gokr/spry
04:33:26	FromGitter	<mratsim> the domain is down so read the blog: http://goran.krampe.se/category/spry/
04:36:45	FromGitter	<mratsim> also for flat buffers: https://github.com/Skrylar/skflatbuffers
04:37:06	FromGitter	<mratsim> there are cap’n proto, protobuf and msgpack lib as well
04:38:32	FromGitter	<mratsim> otherwise for serializing this is quite promising: https://github.com/xomachine/NESM
04:44:57	*	leorize joined #nim
04:54:24	*	BitPuffin joined #nim
04:57:31	*	leorize quit (Quit: WeeChat 2.1)
05:04:01	*	Lord_Nightmare quit (Ping timeout: 248 seconds)
05:04:39	*	Lord_Nightmare joined #nim
05:22:57	FromGitter	<gogolxdong> @shashlick I think walkFiles works. does nimscript module stand alone with c compiling?
05:23:07	*	endragor quit (Remote host closed the connection)
05:32:26	shashlick	Not sure I get that, nimscript is run with nim e
05:32:44	shashlick	It's Nim in script mode, nims file extension
05:32:55	FromGitter	<honewatson> Min Lang https://min-lang.org/
05:33:11	shashlick	Or what you'd use in a nimble file
05:34:48	FromGitter	<gogolxdong> @shashlick yes, it is my concern , we compiled to to c .
05:36:05	shashlick	If you are compiling with nim c then you can use os procs
05:36:30	shashlick	Then there's no nimscript procs available or required for that matter
05:36:39	FromGitter	<gogolxdong> If the boundary of multi-part formdata is ------WebKitFormBoundaryYq9jZBAAoSjRSRKW, how to parse its body?
05:37:13	FromGitter	<gogolxdong> and is this boundary fixed for different browser?
05:38:06	*	xkapastel quit (Quit: Connection closed for inactivity)
05:50:49	*	endragor joined #nim
05:50:51	*	endragor quit (Remote host closed the connection)
05:53:04	*	endragor joined #nim
06:12:03	*	nsf joined #nim
06:12:07	FromGitter	<mratsim> @Araq, here you go with my first test of for-loop macros: https://gist.github.com/mratsim/5fed4b3c1f2f4ea613d155548101a783
06:12:39	FromGitter	<mratsim> It’s promising
06:17:25	*	Senketsu_ joined #nim
06:19:19	Araq	gogolxdong: the boundary is always different.
06:19:37	FromGitter	<gogolxdong> @dom96 how to specify the boundary in multi-part formdata , like WebKitFormBoundaryYq9jZBAAoSjRSRKW, but changes every time.
06:19:47	Araq	you need to store it somewhere and detect it as the sentinel/end value
06:21:35	*	Senketsu_ quit (Ping timeout: 240 seconds)
06:29:40	*	tiorock joined #nim
06:29:40	*	tiorock quit (Changing host)
06:29:40	*	tiorock joined #nim
06:29:40	*	rockcavera quit (Killed (tolkien.freenode.net (Nickname regained by services)))
06:29:40	*	tiorock is now known as rockcavera
06:40:11	*	skelett quit (Ping timeout: 255 seconds)
06:41:24	FromGitter	<Varriount> Araq: Why is it that for procedure ASTs, the pragma is available as a nkPragma node, but it types it's available as an nkPragmaExpr?
06:41:41	FromGitter	<Varriount> *but for type definition ASTs
06:42:00	Araq	{. .} vs identifier{. .}
06:42:32	Araq	an nkPragmaExpr does not contain only the pragma.
06:42:48	FromGitter	<Varriount> Yes, why not make it a sibling instead?
06:43:01	Araq	it's like nkBracket vs nkBracketExpr
06:43:18	Araq	[array construction] vs array[access]
06:44:23	FromGitter	<Varriount> Hrm, I never noticed that you can do `name{.pragma.}` with types. I assumed that the pragma would need to be separated by a space.
06:44:25	Araq	I don't know how it could be a sibling instead
06:45:08	Araq	but it would make some sense...
06:48:08	FromGitter	<Varriount> Araq: I ask because I'm currently refactoring the getCustomPragmaValue macro, as well as some other things.
06:48:31	FromGitter	<Varriount> I was updating the `pragma` getter/setter procs to handle type definitions.
06:49:09	Araq	ok
06:50:48	FromGitter	<Varriount> I did a double take when I realized that pragmas in type definitions aren't direct children of the type definition tree
06:54:31	*	PMunch joined #nim
07:04:30	FromGitter	<ephja> can valueless attributes be specified with buildHtml? not that it's really necessary
07:07:35	*	rokups joined #nim
07:11:03	Araq	br()
07:11:09	Araq	^ like so?
07:11:20	*	couven92 joined #nim
07:17:48	FromGitter	<ephja> attributes, not elements; but <input disabled="..."> works
07:20:11	FromGitter	<mratsim> Regarding pragma, the {.inheritable.} pragma is really weird most type pragma are on the left but inheritable is on the right `type Foo = object {.inheritable.}`
07:24:24	FromGitter	<Varriount> @mratsim Really? Wow.
07:24:34	*	skelett joined #nim
07:25:11	FromGitter	<ephja> "buildHtml..." gah, so much typing. just make every element a macro so we don't have to type that every time ;)
07:25:32	FromGitter	<Varriount> ephja: PR?
07:25:54	*	BitPuffin quit (Remote host closed the connection)
07:26:10	*	user1101 joined #nim
07:26:20	FromGitter	<ephja> great lib
07:27:07	FromGitter	<Varriount> I mean, you could make a pull request adding the element macros
07:30:25	*	Lord_Nightmare2 joined #nim
07:31:13	*	Lord_Nightmare quit (Ping timeout: 248 seconds)
07:31:18	*	Lord_Nightmare2 is now known as Lord_Nightmare
07:32:25	*	BitPuffin joined #nim
07:35:46	*	Arrrr joined #nim
07:35:46	*	Arrrr quit (Changing host)
07:35:46	*	Arrrr joined #nim
07:43:15	FromGitter	<alehander42> @ephja you could just write your own `h` alias for buildHtml
07:44:19	FromGitter	<alehander42> I certainly wouldn't want my namespace polluted with span, br head and all of those
07:45:19	FromGitter	<mratsim> ^+1
07:45:59	Araq	I'm really happy with Karax's DSL and don't want to change it
07:46:50	FromGitter	<ephja> more pollution!
07:47:37	FromGitter	<Varriount> Hm, you could do something like an `import htmlElements` in the procedure itself.
07:48:01	FromGitter	<Varriount> (Assuming Nim supports non-global imports)
07:48:04	FromGitter	<alehander42> yeah but that would be way more writing than just buildHtml :D
07:48:32	Araq	we don't support non-global imports.
07:48:51	Araq	they mess up the scoping rules, D had to add special rules for these
07:54:29	*	skrylar joined #nim
08:02:53	*	Lord_Nightmare2 joined #nim
08:04:17	*	Lord_Nightmare quit (Ping timeout: 248 seconds)
08:04:29	*	yglukhov quit (Read error: Connection reset by peer)
08:05:16	*	yglukhov joined #nim
08:07:53	*	Lord_Nightmare joined #nim
08:08:45	*	Lord_Nightmare2 quit (Ping timeout: 264 seconds)
08:12:54	*	Lord_Nightmare2 joined #nim
08:14:24	*	Lord_Nig- joined #nim
08:16:33	*	Lord_Nightmare quit (Ping timeout: 265 seconds)
08:16:34	*	Lord_Nig- is now known as Lord_Nightmare
08:17:45	*	Lord_Nightmare2 quit (Ping timeout: 264 seconds)
08:20:03	*	BitPuffin quit (Remote host closed the connection)
08:23:18	*	BitPuffin joined #nim
08:29:02	*	Senketsu_ joined #nim
08:30:11	FromGitter	<narimiran> @mratsim no love for `forAll`? :(
08:31:22	FromGitter	<Varriount> @krux02 Any chance your AST pattern matching library will support partial matching?
08:31:54	FromGitter	<Varriount> I often need to match only the start of a tree, or from the end
08:32:39	*	Senketsu_ quit (Client Quit)
08:35:23	*	Senketsu_ joined #nim
08:35:59	FromGitter	<mratsim> @narimiran Anyway when Araq’s for-loop macro are out loopfusion will be either deprecated or rewritten or integratted in sugar.nim I’d say.
08:37:16	FromGitter	<mratsim> at least forZip can be found when we search for “Zip"
08:37:33	FromGitter	<mratsim> though it doesn’t roll off the tongue as nicely as forAll
08:41:59	FromGitter	<narimiran> ok, i get the rationale, hopefully we'll get built-in for-expresions soon....
08:47:19	Araq	don't hold your breath
08:47:42	Araq	on the other hand, nothing in for-loop macro's design is tied to statements...
08:52:24	*	dddddd joined #nim
08:54:57	*	skrylar quit (Remote host closed the connection)
08:55:09	*	skrylar joined #nim
09:05:45	FromGitter	<mratsim> @alehander42 I checked your quicktest lib, nice work. I’d like more documentation on the number of tests, I’d like to do a 1 million, not 20 :P. Also it would be cleaner to use slices instead of min=0, max=20.
09:07:14	*	Senketsu_ quit (Quit: WeeChat 2.1)
09:07:45	FromGitter	<alehander42> yeah, the slice thing can be very easily arranged
09:08:24	FromGitter	<alehander42> I think right now you can already pass ⏎ ⏎ ```quicktest name, count do (arg: constraint): ⏎ ..``` [https://gitter.im/nim-lang/Nim?at=5adef40862316e0505f34c83]
09:08:30	FromGitter	<alehander42> the count
09:09:08	*	Senketsu_ joined #nim
09:09:24	FromGitter	<alehander42> the only thing I am not sure is if there wouldn't be a problem with the <x>int<y> types, can't remember if I added default handlers for them
09:09:33	FromGitter	<mratsim> 1_000_000 in non-release mode is a bad idea™
09:09:43	FromGitter	<mratsim> Unhandled exception: /dev/urandom is not available
09:09:45	FromGitter	<mratsim> lol
09:10:23	FromGitter	<alehander42> well, yeah and I've tested it on linux and the js backend
09:10:31	FromGitter	<alehander42> for c backend the random gen depends on
09:10:52	FromGitter	<mratsim> na it works but it crashed at one point
09:14:46	FromGitter	<alehander42> hm, weird, I thought `/dev/urandom` should be always available
09:15:04	FromGitter	<alehander42> ah, you already opened an issue for uint, ok :D
09:17:35	*	Senketsu_ quit (Quit: WeeChat 2.1)
09:18:27	FromGitter	<alehander42> ah yeah, `min` and `max` are indeed useful, as Nim doesn't have infinite ranges
09:18:38	FromGitter	<alehander42> so you couldn't just specify e.g. -inf..4
09:18:49	FromGitter	<alehander42> but an additional range arg would be still good
09:35:54	*	cspar quit (Ping timeout: 255 seconds)
09:37:45	*	gangstacat quit (Quit: Ĝis!)
09:45:47	*	gangstacat joined #nim
09:55:48	Araq	wow escape is used quite a lot
09:58:01	FromGitter	<mratsim> also @alehander42 it would be nice if it only prints when there is a failed test.
09:58:55	FromGitter	<mratsim> and then nice to have, having a counter 133/1000 tests (but that’s really like not important for quick tests)
10:00:30	FromGitter	<mratsim> ah it printed everything because of the urandom crash
10:00:33	FromGitter	<mratsim> :/
10:00:35	FromGitter	<arnetheduck> @mratsim see - that's why those kind of things shouldn't be exceptions, but part of the API of the function that gives a random number, such that you explicitly ignore it ;) the "90% errors are passed up" isn't because it makes sense to pass errors up 90% of the cases - it's because exceptions encourage people not to think about how things fail, and write shitty code like that...
10:01:54	Araq	arnetheduck: cost-benefit analysis, it's a thing.
10:02:00	FromGitter	<mratsim> urandom not being available is exceptional though and
10:02:19	Araq	do a decent one to convince me.
10:02:24	FromGitter	<arnetheduck> not really.. it's just part of life
10:02:49	FromGitter	<mratsim> crashing is a good thing. You don’t want to produce sensitive random numbers with a broken urandom (though you shouldn’t anyway)
10:03:29	FromGitter	<arnetheduck> I think it's even a better thing if the api tells you it may go wrong and you choose whether to crash or not
10:04:12	Araq	you can check any large Go project to prove or disprove this 90% number.
10:04:43	FromGitter	<mratsim> I already see all Nim function with the {.discardable.} pragma :P
10:04:49	FromGitter	<arnetheduck> you don't think google did that analysis?
10:05:11	Araq	yes, I don't think so, Go was development based on gut feelings.
10:05:15	FromGitter	<alehander42> don't cargo cult google
10:05:16	FromGitter	<mratsim> Na, Google did a language they can teach to fresh out of school graduates
10:05:25	FromGitter	<alehander42> rob pike's team is very opionated
10:05:35	codevantage	mratsim i just use void instead
10:05:39	FromGitter	<mratsim> Araq is very opinionated too ;)
10:05:49	FromGitter	<arnetheduck> yeah, that's true.. they set out with a bunch of principles more likely and went from there
10:06:05	FromGitter	<alehander42> also this has philosphically nothing to do with errors vs exceptions but with how much is error checking enforced: ⏎ you can theoretically have strict exception tag checking and also ⏎ you can easily have error-based system where it's easy to ignore
10:06:30	FromGitter	<alehander42> this includes go too: iirc if your function returns just an error (it's void otherwise) you can actually just write `f()`
10:06:53	FromGitter	<alehander42> which is totally the opposite of the whole "you have to `(value, .., err) = call()`
10:07:10	FromGitter	<mratsim> well the thing is should 100’i8 + 100’i8 throw an exception or return an error code?
10:07:41	FromGitter	<arnetheduck> regardless, that's all beside the point I was making - which is that the choice of error model deeply affects the way that people write programs
10:08:23	FromGitter	<alehander42> the *nuances* in the particular design/implementation of an error model deeply affects how people write programs
10:08:54	Araq	your point might as well be that {.raises: [].} is heavily under-used
10:09:00	FromGitter	<arnetheduck> by hiding the fact that errors exist, you incentivize people to ignore errors - does anyone think that's controversial thing to claim
10:09:33	Araq	no, it's true. it's also a question of software development costs.
10:09:50	FromGitter	<arnetheduck> and as a corollary to that, if the compiler doesn't tell you, you'll ignore it
10:10:06	FromGitter	<alehander42> all I am saying is , you can still easily pretend errors don't exist in some error-based systems and you can have exception-based systems where you can't ignore possible errors
10:10:12	FromGitter	<mratsim> basically one of the main selling point of Arraymancer is being able to do: `let y = a * x + b` instead of `y = tf.add(tf.mul(a, x), b)` which is even worse if Google choosed to use error code.
10:10:19	*	Senketsu_ joined #nim
10:10:21	FromGitter	<alehander42> it is not an intristic property of exception:error at all
10:10:48	Araq	^ alehander42 raises a good point.
10:10:57	FromGitter	<mratsim> also I think you can use a macro/const that enforce that all exceptiosn that can be thrown are checked exhaustively
10:11:25	FromGitter	<arnetheduck> @mratsim you can, but the language and the std lib sets the tone
10:11:57	FromGitter	<alehander42> also, you can still have most of both worlds: a block where you make calls (maybe using `?` to signify possible errors) and after that a compiler-enforced handler to deal with possible errors *or* escalate them up explicitly
10:12:28	FromGitter	<mratsim> I think @arnetheduck point is that if the stdlib uses exceptions, it’s hard to build on top on that when you require no exceptions
10:12:59	FromGitter	<mratsim> but the same is true for errors, I don’t know of a better system though (design by contract?)
10:13:06	FromGitter	<arnetheduck> @alehander42 true. I'm mostly arguing from the current reality of nim where errors are hidden, and I'm working on a project that will be managing your money
10:13:12	*	endragor_ joined #nim
10:13:26	FromGitter	<alehander42> it all depends: the stlid can tag all of its exceptions and you can have a way to call a function which turns the exception into an error
10:13:55	Arrrr	Really?
10:14:17	FromGitter	<alehander42> I mean that would be one possible implementation of a language where both can coexist
10:15:30	*	endragor quit (Ping timeout: 255 seconds)
10:16:30	Arrrr	I'd remove some of the exceptions in the std library, starting with the ones that happens while handling sockects. I'd rather check the result than instantiate an exception and face the cost of raising it
10:17:14	FromGitter	<mratsim> exceptions should be kept to exceptional stuff, a network connection can fail.
10:17:28	FromGitter	<arnetheduck> so can `/dev/urandom`, arguably
10:18:06	Araq	now we're getting somewhere; removing exceptions from the socket handling is something I can sympathize with
10:18:20	Arrrr	But I have always the choice to wrap whatever function inside a try-except.
10:18:24	Arrrr	*can
10:18:29	FromGitter	<mratsim> yes, but it should throw, it’s unsafe not to throw as an attacker can block /dev/urandom and it keeps giving you the same number otherwise (assuming it doesn’t throw)
10:18:54	Araq	on the other hand, rewrite osproc.nim in the style you propose and see where it takes you
10:19:23	Araq	the code just becomes check posixCall(....)
10:19:25	FromGitter	<arnetheduck> @mratsim it's unsafe to pretend you got a random number when you didn't.. you don't need an exception for that though
10:19:25	Araq	instead of
10:19:31	Araq	posixCall(...)
10:19:46	Araq	it just adds noise.
10:19:50	*	Senketsu_ quit (Quit: WeeChat 2.1)
10:19:56	FromGitter	<mratsim> but what should urandom do instead, it can’t return an error code
10:20:20	FromGitter	<arnetheduck> it can return an optional value
10:20:44	FromGitter	<arnetheduck> forcing you, the caller of the api, to make a conscious choice
10:21:38	*	Senketsu_ joined #nim
10:21:41	FromGitter	<arnetheduck> Araq, and for reliable software, it's not noise to say that you're ignoring errors explicitly - same as `discardable`
10:21:53	FromGitter	<alehander42> just wondering, why would enforcing the caller to check for the call's possible exceptions described in `{.raises: }` be any different?
10:22:16	FromGitter	<arnetheduck> so `check: posixcall1(); posixcall2()` is quite reasonable, overhead-wise
10:22:46	Araq	well Nim is so good at removing boilerplate that it muddles the arguments :-)
10:23:38	FromGitter	<krux02> Araq: did you see the matchAstRecursive construct?
10:23:42	*	Senketsu_ quit (Client Quit)
10:24:19	*	Vladar joined #nim
10:24:33	Araq	but still. Let's analyse some large Go programs.
10:24:56	FromGitter	<krux02> After you mentioned it yesterday evening I just implemented it.
10:25:05	Araq	krux02: link please
10:25:11	FromGitter	<arnetheduck> @alehander42 it wouldn't be different, but it would also not be feasible or useful - anything that does arithmetic throws so there's practically nothing you can do without exceptions
10:25:55	FromGitter	<krux02> https://github.com/krux02/ast-pattern-matching/blob/master/src/ast_pattern_matching.nim#L503
10:27:47	Araq	exceptions are hurting me in the compiler btw.
10:27:54	Araq	why?
10:28:21	Araq	because I shouldn't do "on error bubble up".
10:28:53	Araq	instead semantic checking should continue and do its best. it's required for nimsuggest to work.
10:29:02	FromGitter	<alehander42> @arnetheduck so, sometimes it's ok to ignore errors, but sometimes not ?
10:29:35	Araq	so it's not like I can't see the merit in your argumentation.
10:29:51	FromGitter	<alehander42> how would you deal with arithmetic edge cases then in an error-based system?
10:30:03	*	SenasOzys joined #nim
10:30:28	Araq	alehander42: with saturated arithmetic, preferably one where min/max are sticky. like NaN is.
10:30:29	FromGitter	<krux02> Araq: I thought about your argument about range and that you want to deprecate static[T]
10:30:30	*	SenasOzys__ quit (Read error: Connection reset by peer)
10:30:39	FromGitter	<alehander42> @arthe with exceptions, it would be simple: a call can still say : "this exceptions should be checked
10:30:57	FromGitter	<krux02> I came to the conclusion that the range type is nothing else than a special case of static[T]
10:30:59	FromGitter	<alehander42> and these exceptions can be ignored"
10:31:17	FromGitter	<krux02> it is static[tuple[low,high:int]]
10:31:19	FromGitter	<arnetheduck> @alehander42 they tried that in c++ and removed it.. they also tried it in java, and everyone started using the unchecked ones
10:31:56	FromGitter	<alehander42> @arnetheduck so why did that happen? because people don't like to check all of their errors
10:32:42	FromGitter	<krux02> @alehander42 please no, checked exceptions are awful
10:32:48	FromGitter	<alehander42> still, you have to give an alternative: how would you deal with a `/` that can fail
10:33:24	FromGitter	<mratsim> it returns (div, mod, error code)
10:33:26	FromGitter	<krux02> Java has checked exceptions and it leads to catch-discard handling to silence tho compiler. That just makes the code worse
10:33:38	FromGitter	<alehander42> @krux02 but some error-based systems seem to me as bizarre+more manual based half broken implementation of checked exceptions
10:33:41	FromGitter	<alehander42> e.g. Go
10:33:58	*	max3 quit (Quit: Connection closed for inactivity)
10:34:30	FromGitter	<krux02> and Scala (compiles to java virtual machine) understands java exceptions, can declare some "throws" annotations, but generally it just ignores all of the throws declarations
10:34:49	FromGitter	<alehander42> @krux02 that's not a fundamental issue with them, you can do the same in Go in two different ways
10:35:10	FromGitter	<krux02> the go error result type isn't cool either
10:35:32	FromGitter	<alehander42> @mratsim so how do you chain `/` then?
10:35:41	FromGitter	<krux02> and it is not mandatory.
10:35:49	FromGitter	<mratsim> you can’t and it’s sad
10:35:59	Araq	alehander42: I just told you. saturated arithmetic.
10:36:21	FromGitter	<mratsim> how do you saturate a divide by zero?
10:36:31	Araq	that's just MAXINT
10:36:56	FromGitter	<mratsim> 0/0 is not maxint
10:36:56	Araq	the model I propose is widely deployed in the form of IEEE floating point math.
10:36:58	FromGitter	<alehander42> @Araq I am talking in general
10:37:08	Arrrr	I'd rather have the option of throwing exceptions than nothing at all. It is not a matter of using exceptions or return code for everything.
10:37:09	FromGitter	<alehander42> about operators and various problem domains
10:37:28	FromGitter	<alehander42> it's always possible that an operation would make sense to have an exceptional condition
10:37:41	Araq	in general you need NaN or something comparable.
10:38:12	FromGitter	<krux02> @alehander42 yea but the exceptional conditions should be assumend to be exeptional and therefore should not be forced to be checked
10:38:36	FromGitter	<krux02> I think the best alternative to excepitons is the result type from rust.
10:39:03	Araq	I think it doesn't work. Rust dies on OOM.
10:39:17	FromGitter	<alehander42> @krux02 I think that's just a not very useful language argument, exceptions are one error-handling mechanism, the fact that they are called "exceptions" shouldn't matter
10:39:32	FromGitter	<krux02> OOM?
10:39:43	Araq	out of memory.
10:40:09	Araq	if something like Result would truely work it would be useful for basic arithmetic and OOMs.
10:40:13	FromGitter	<alehander42> @Araq yes but NaN is not an error: you can't return info about *what* happened that was wrong
10:40:31	FromGitter	<krux02> well almost all programs die on out of memory. But out of memory isn't really happening anymore, it is rather the entire system is swapping and becomes unusable.
10:40:54	Araq	that's not the point.
10:42:43	FromGitter	<krux02> then I don't get the point.
10:42:58	FromGitter	<krux02> But I also think it does really matter at the moment
10:43:01	Araq	the point is OOM and arithmetic errors are really pervasive and so nobody uses Result<> for it.
10:43:35	FromGitter	<krux02> ah, ok, now I get it
10:44:04	Araq	there is a reason Haskell, Ada, Erlang all have exceptions.
10:44:26	Araq	and two of these have been designed with incredible reliability in mind.
10:44:50	Araq	do you really think the Haskell people don't know about the Either monad?
10:48:56	Araq	btw another argument against exceptions would be that while the "stack unrolling" is automatic there is no "heap unrolling" and usually the heap/stack distinction is a low level detail programmers are not aware of.
10:49:47	*	simonsolnes joined #nim
10:49:47	Araq	so yeah, we can enumerate the arguments, let's look at some numbers.
10:51:14	FromGitter	<krux02> I know about the either monad from scala, and scala has a lot of Haskell influence.
10:51:35	FromGitter	<krux02> But I didn't find the Either Monad very usable
10:53:11	FromGitter	<krux02> Araq: I am very aware if something is on the heap or the stack.
10:54:19	FromGitter	<krux02> But we should all agree that exeption handling is not a pattern that should be sprinkled all over the codebase.
10:55:22	codevantage	core code base or in general?
10:56:34	FromGitter	<krux02> in general
10:56:46	FromGitter	<krux02> I don't know what you mean with core code base
10:56:58	codevantage	like nim itself
10:57:05	codevantage	kernel
10:57:22	codevantage	the nim stdlib
10:57:32	FromGitter	<krux02> execption handling is somthing that should not be in the core at all. Maybe somewhere outside in the outer shell parts of the program to do error reporting
10:57:37	FromGitter	<ephja> how do I run some code after karax has rendered the DOM? that is, some time after onload has been dispatched
10:58:05	FromGitter	<krux02> what do you like about the nim stdlib?
10:59:14	codevantage	stdlib works.
10:59:48	Araq	ephja: there is a onPostRender callback you can pass to karax's init
10:59:54	FromGitter	<krux02> works?
10:59:56	Araq	sorry, forgot the real names.
11:00:24	FromGitter	<krux02> Araq: you still didn't loose a word about the recursive ast pattern matching that I wrote because you asked for it
11:00:40	Araq	krux02: oh yeah, how can it work?
11:01:02	codevantage	as long as stdlib works then what else can i ask for?
11:01:05	Araq	how did you solve the problem that you don't know the 'self(paramsHere)' call?
11:01:24	FromGitter	<krux02> it just does the normal matching on the root node and when it doesn't match it does also on all child nodes
11:01:49	FromGitter	<krux02> which self?
11:01:49	Araq	but it needs to do it recursively
11:01:54	Araq	exactly.
11:02:35	FromGitter	<arnetheduck> I'd think performance numbers are largely uninteresting in this discussion and perhaps useful as the last tipping of the scale, all else being equal .. even the whole zero-cost vs if-branches vs EH tables is in the zone of negligible things that you can optimize where it matters.. what I'm missing is that when you're working with exceptions, you can come pretty far without caring about errors and you have this warm
11:02:35	FromGitter	... chummy feeling everything is fine, because your understanding of what can go wrong is inherently limited to your own little head, then you end up with a shitty code base that you wish dealt with errors across the board in a sane way (ie compiler, nimsuggest).. there's no teaching moment where you can draw upon the ... [https://gitter.im/nim-lang/Nim?at=5adf0ecb1130fe3d360e8405]
11:02:38	FromGitter	<krux02> I wrote a function that calls a callback recursively on each node
11:03:03	FromGitter	<krux02> the result of the function is a bool and says it the function should go down deeper or not.
11:04:00	FromGitter	<krux02> and the matching expression just compilies to a call to this function.
11:04:03	FromGitter	<arnetheduck> it's like globals.. sure, they're a nice shortcut, but when was it that you had a nicely designed lib and thought to yourself, "oh, I really wish I had used a global here"? the reverse happens almost every time a mutable global is involved, usually at a point when it's expensive to do something about it
11:05:42	FromGitter	<krux02> @arnetheduck well sometimes globals are just the better alternative
11:06:15	FromGitter	<krux02> I know mutable global variables suck, but sometimes that is what suits the purpose best.
11:06:21	FromGitter	<arnetheduck> yeah, sure, as long as you're writing a 1-page script, they're great ;)
11:06:40	FromGitter	<krux02> on I mean really big projects
11:06:49	FromGitter	<krux02> sometimes it is just not suitable to do anything else
11:07:26	FromGitter	<alehander42> @arnetheduck again, this is true, but it has nothing to do *fundamentally* with the fact that a system is exception or error-based, but with the way it enforced one to deal with those cases
11:08:16	FromGitter	<arnetheduck> and again, @alehander42 , I'm looking at exceptions as they look and are used in nim, today, after 10 years of std lib development and whatnot
11:08:29	FromGitter	<alehander42> oh exactly, that's my next point
11:08:50	FromGitter	<arnetheduck> and clearly, something is wrong there, and I don't want my next project to end up the same ;)
11:08:53	FromGitter	<alehander42> if I have an existing language & ecosystem, which would be easier: rewrite 80% of error handling *everywhere* into a different paradigm, or improve the existing one, so people couldn't ignore it so easily
11:09:25	Araq	well you can't blame exceptions for every problem in Nim.
11:10:03	*	athenot joined #nim
11:10:45	codevantage	my main problem with exceptions is that people dont use them, or think printing a string is an exception?
11:10:59	Arrrr	Btw, how are symbol files going on?
11:11:37	FromGitter	<arnetheduck> no, of course not, there are many kinds of evils that experience has taught us could be improved :) I used to love exceptions!
11:12:07	Araq	oh that's interesting to hear.
11:13:06	FromGitter	<arnetheduck> .. and I still like them for some things!
11:13:09	*	athenot_ joined #nim
11:13:23	Araq	Arrrr: they are like the dragons in game of thrones.
11:13:30	Araq	they are coming.
11:13:50	FromGitter	<krux02> Wasn't that the winter
11:14:09	FromGitter	<arnetheduck> @krux02 what are those sane usages of writable globals?
11:14:27	Arrrr	I haven't read GoT
11:14:50	FromGitter	<krux02> @arnetheduck http://etodd.io/2017/03/29/thirteen-years-of-bad-game-code/
11:14:57	*	athenot quit (Ping timeout: 240 seconds)
11:15:04	Araq	arnetheduck: it depends on the problem domain. for games you have this huge messy mutable shared state and passing it to every proc via a parameter doesn't make it prettier.
11:15:09	FromGitter	<alehander42> so yeah, replacing exceptions before v1 is an insane idea, except if you're ready to rewrite 100_000 nim lines of code
11:15:20	FromGitter	<krux02> he mentions the World Object that he passes down to everything
11:15:21	FromGitter	<alehander42> improving them on the other hand: entirely possible
11:15:41	FromGitter	<krux02> but the program doesn't have several World instances, and the World is just mutable
11:15:57	FromGitter	<krux02> it is just much more sane to keep the world a global thing
11:17:45	*	Lord_Nightmare quit (Ping timeout: 264 seconds)
11:18:04	*	flyx quit (Ping timeout: 260 seconds)
11:18:20	FromGitter	<krux02> "However, almost all gameplay code needs access to the entire world state. I 'solved' this problem by storing everything in a 'world' object and passed the world into every single function. No more globals! I thought this was great because I could theoretically run multiple, separate worlds simultaneously."
11:19:51	FromGitter	<arnetheduck> @krux02 from what I can tell, that article doesn't say anything about globals vs non-globals.. it merely states that if you have a design that includes a global, make it a global and not a disguised local
11:19:53	FromGitter	<krux02> @arnetheduck I am completely on your side when you mean avoidable globals. Or avoidable global state. But obscuring global state, well that is just bad.
11:20:28	*	noonien joined #nim
11:20:44	FromGitter	<krux02> @arnetheduck I just gave you the quote.
11:21:02	Araq	well in the compiler the disguised locals are much better than real globals... :-)
11:21:16	FromGitter	<krux02> Araq: are they?
11:21:20	Araq	yep.
11:21:31	Araq	I am slowing getting rid of the globals.
11:21:45	FromGitter	<arnetheduck> in the compiler, a main reason I don't wanna touch repl for nlvm is all the globals.. once you're done, I might consider it :)
11:21:47	FromGitter	<krux02> if it makes you happy.
11:22:18	Araq	globals make the compiler-as-an-API super shitty.
11:22:53	Araq	doesn't apply to game development though, games are usually not shipped as an API.
11:23:06	FromGitter	<krux02> modding API :P
11:23:09	*	Lord_Nightmare joined #nim
11:23:33	Araq	the globals also make symbol files harder...
11:24:09	FromGitter	<krux02> Well I can't really argue here, because I have never written a real compiler
11:26:55	FromGitter	<alehander42> basically if somebody will reuse your code as a lib somewhere, globals are bad, otherwise you probably can get away with a `world` object if you really use one
11:27:16	FromGitter	<mratsim> I think the guys who wrote OpenCL forgot about that
11:28:11	FromGitter	<alehander42> I guess it shouldn't be too hard to also refactor such a thing automatically: if you need, just gen the static call graph, everywhere you depend (or your calls) on the `world` object, pass it as first argument
11:28:16	FromGitter	<mratsim> You basically have to have a global or ask everyone to call the kernel with `let err = OpenCL_enqueue_write(context, x, y, awesome_add_kernel)`
11:28:36	codevantage	theres a diff between module level global and magical globals
11:29:33	Araq	what are magical globals?
11:29:56	FromGitter	<mratsim> also error codes mean “let” variable becomes incredibly painful to use.
11:32:19	codevantage	magical globals is something where you import a module and it effects modules that havent imported it
11:33:30	codevantage	but if you already know what globals a module holds there isnt any surprise.
11:37:06	*	flyx joined #nim
11:40:35	FromGitter	<mratsim> I like this pieace on Exception vs Errors: https://nedbatchelder.com/text/exceptions-vs-status.html
11:44:52	FromGitter	<mratsim> the comments section show that the debate was still as hot 15 years ago
11:46:03	FromGitter	<alehander42> the invisible code argument is good one, haven't thought of it
11:52:11	FromGitter	<Vindaar> One thing about status codes I don't get: if I want to have a proc return some value, I can't return a status code as well (unless I always return a tuple). Or I have to use procs with a `var` arg instead, which is mutated (esp. in Nim when I love to use `let` that's a deal breaker). Both are ugly to me. Am I missing something?
11:53:18	FromGitter	<krux02> aren't status codes just for unix shell
11:53:45	FromGitter	<Vindaar> we're talking in context of exceptions / error handling, no?
11:53:52	FromGitter	<krux02> ah, ok
11:54:04	FromGitter	<krux02> status codes are hidden error values.
11:54:13	FromGitter	<krux02> just return the error, status codes stink
11:54:53	FromGitter	<Vindaar> wait, "just return the error", what do you mean? Maybe I don't get your distinction between status code and error
11:55:43	FromGitter	<alehander42> @mratsim I have support for many ints and ranges now, the urandom thing is very weird on my machine: it seems if I tweak the random lib to get bigger chunks from urandom, it can easily support indefinite amount of values, but if you do >1022 of OPEN /dev/urandom it
11:55:45	FromGitter	<alehander42> oh wait
11:56:03	FromGitter	<alehander42> oh man
11:56:13	FromGitter	<alehander42> ok I'll fix the random lib
11:56:22	FromGitter	<alehander42> it opens >=1024 connections probly
11:57:01	FromGitter	<mratsim> there’s an additional “real example” for the debate between this dev and Joel Spolsky (founder of Stack Overflow): https://nedbatchelder.com/text/exceptions-in-the-rainforest.html
11:57:08	FromGitter	<krux02> an error is an object or a pointer to an object that tells you what is wrong. A status code in an idex into an unknown array of predefined error objects, where you first have to find out what array is actually meant where that array is to know what's wrong
11:57:36	FromGitter	<Vindaar> @krux02 ahh, that makes sense. thanks
11:57:47	FromGitter	<arnetheduck> @mratsim I think everyone agrees that using plain error codes is broken in all the ways you say - no way to compose, monopolizes return code, leads to quirky API, and something else was systematically introduced, it wouldn't be that
11:58:09	FromGitter	<mratsim> The challenge in building a large system is making sure errors get communicated around. Exceptions are a better way to do that than status returns: ⏎ ⏎ Exceptions can carry richer information. If error handling is so important, why try to cram it all into a DWORD? ⏎ Exceptions let the B-layer get on with its work without being a mindless bucket brigade for status returns. ⏎ Exceptions make human error (failure
11:58:09	FromGitter	... to catch) visible, while error returns make human error (failure to check) invisible. ... [https://gitter.im/nim-lang/Nim?at=5adf1bd126a769820b2bfb1c]
11:58:45	FromGitter	<mratsim> Well in Nim we would have Error Enum not error code.
12:00:42	FromGitter	<mratsim> I’m all for something better than exceptions as long as it does not clutter function return values.
12:01:24	FromGitter	<alehander42> wait this bug is fixed in the lib
12:05:53	*	SenasOzys quit (Ping timeout: 248 seconds)
12:06:14	FromGitter	<arnetheduck> @mratsim using `rust` results as an example, ⏎ ⏎ > Exceptions can carry richer information. If error handling is so important, why try to cram it all into a DWORD? ⏎ ⏎ so do `results`, if you want them to - canonical example here is the `Future` - you raise an error, catch it, store it in a future instance, read it from the instance and then rethrow it so there's no reason a result cannot do the same
12:06:14	FromGitter	... ... [https://gitter.im/nim-lang/Nim?at=5adf1db62b9dfdbc3ac30774]
12:07:11	FromGitter	<alehander42> @dom96 if a lib has a new commit after tag x, I can only install it with `lib #head` right?
12:07:21	FromGitter	<alehander42> that was my mistake, the lib is fixed, it just needs a new version
12:07:53	FromGitter	<alehander42> tag
12:11:05	FromGitter	<mratsim> @arnetheduck I agree that Results/Either are better than error codes. They really need synctatic sugar (no `unwrap`, but `>>=` or `?` are fine). The thing is, if something fails in the low-level part of your code, and you want to display it on CLI or in a UI, you have to bubble up the error to the top. Meaning all layers should be have some stuff to carry low-level failures.
12:11:25	FromGitter	<mratsim> I’m not sure how to do that cleanly?
12:13:10	*	ieatnerds joined #nim
12:15:09	*	SenasOzys joined #nim
12:16:20	FromGitter	<arnetheduck> if you want to display something on a high-quality UX, for sure it's not gonna come as a low-level exception from the bowels of your app.. to make it interesting, let's use a regex as an example, and what the choices signal: ⏎ ⏎ 1) an exception tells you that malformed regexes are really weird and the user should be punished with a call stack for making them.. the maker of the GUI will likely not catch the
12:16:20	FromGitter	... exception and deal with it, except in some high-level catch-all and ugly generic dialog that follows
12:16:48	bodie_	any pointers on gui? I was looking at the nuklear bindings but I'm hesitant
12:17:01	FromGitter	<arnetheduck> 1) an error code will lead to the developer of the UX to consider both as things that can naturally happen, and will choose to shade the regex edit box red
12:17:22	*	alpha joined #nim
12:17:45	*	alpha is now known as Guest56261
12:18:22	FromGitter	<arnetheduck> yet, when regexes are involved, I can be pretty sure that languages with exceptions will have a regex library that raises on syntax error - that's the most natural thing to do
12:18:42	Guest56261	Re: exceptions vs errors - zig (another new WIP language) also NOT uses exceptions, it uses something similar to Rust
12:18:44	Guest56261	https://ziglang.org/documentation/master/#Errors
12:19:03	bodie_	C also doesn't use exceptions. :P
12:19:09	bodie_	You can bolt them on though.
12:19:39	FromGitter	<mratsim> you can do a try finally though
12:19:40	Guest56261	Well, sane people nowadays also don't use C unless they work on a problem domain where they don't have a choice.
12:20:53	*	Guest56261 is now known as Alpha1220
12:23:57	*	euantor quit ()
12:24:02	bodie_	sane people also don't use exceptions: https://google.github.io/styleguide/cppguide.html#Exceptions
12:24:15	*	euantor joined #nim
12:24:33	Araq	well the things is, Nim is not Java or C# or C or C++.
12:24:51	Araq	Nim's exceptions have been enriched with exception tracking.
12:24:56	FromGitter	<arnetheduck> bodie_ that guide is a sad lobotomization of c++ :)
12:25:16	Araq	we need real world data on how .raises: [] works in practice.
12:25:55	*	leorize joined #nim
12:26:27	*	Senketsu_ joined #nim
12:26:29	FromGitter	<Vindaar> @bodie_: well... "Our advice against using exceptions is not predicated on philosophical or moral grounds, but practical ones. Because we'd like to use our open-source projects at Google and it's difficult to do so if those projects use exceptions, we need to advise against exceptions in Google open-source projects as well. Things would probably be different if we had to do it all over again from scratch."
12:26:32	Araq	C started with no exceptions, C++ added them, Java made them checked, C# made them unchecked again
12:26:50	Araq	Nim made them tracked.
12:27:49	Araq	the idea here was that you start with quirky code and gracefully add .raises: [] into some critical key points to make the compiler prove it is safe.
12:27:53	FromGitter	<arnetheduck> Araq, well, it doesn't, if 10 years of nim experience is any guide - if it was a killer feature, it would be all over the place
12:28:29	Alpha1220	Most people won't use .raises:[] or any other safety precautions if they have any friction associated with them, most people wan't to prototype quickly. Nim is really great at that IMHO but well if you think about errors or exception tracking or whatever, it is better to enforce security than to make stuff optional.
12:28:29	Araq	sorry, but 10 years of Nim experience really isn't any guide. ;-)
12:29:08	Araq	Swift added exceptions too afaik.
12:29:16	bodie_	exceptions / panics are a last-ditch resort to avoid undefined behavior I suppose.
12:29:28	Araq	and is recently new and Objective-C wasn't big on exceptions either.
12:29:42	bodie_	wouldn't it be better just to disallow that and provide a macro to make raising errors easier? something like Rust's ? macro
12:30:20	bodie_	quick question here: does Nim support MSVC? https://nim-lang.org/install_windows.html
12:30:43	bodie_	I'm not comfortable using GCC / MinGW for Windows-native applications
12:30:58	Alpha1220	https://developer.apple.com/library/content/documentation/Swift/Conceptual/Swift_Programming_Language/ErrorHandling.html Exception handling without call stack unwinding.
12:31:01	FromGitter	<mratsim> You can compile it with VCC
12:31:52	bodie_	mratsim -- I don't want my users on Windows to have any difficulty at all. if I can just toss them a tool and a source, it should just work as long as they have msvc installed, imo. it's hard enough to get windows users to set up msvc and the developer prompt
12:33:21	Alpha1220	Nim works with msvc I think (except for vs2017 because it changed the paths and there is a WIP since a long time PR (and a few Issues on GH)) regarding vs2017.
12:34:05	bodie_	like, actual work in actual progress, or "dead end" :P
12:34:39	Araq	Alpha1220: these look like exceptions to me
12:35:51	*	odc quit ()
12:35:57	Araq	"Because the vend(itemNamed:) method propagates any errors it throws, any code that calls this method must either handle the errors—using a do-catch statement, try?, or try!—or continue to propagate them. For example, the buyFavoriteSnack(person:vendingMachine:) in the example below is also a throwing function, and any errors that the vend(itemNamed:) method throws will propagate up to the point where the buyFavoriteSnack(person:vendingMachine:
12:35:57	Araq	) function is called."
12:36:09	*	odc joined #nim
12:36:10	Alpha1220	yep, but it is written over there that
12:36:23	Araq	bodie_: unzip and run finish.exe
12:36:33	Alpha1220	NOTE Error handling in Swift resembles exception handling in other languages, with the use of the try, catch and throw keywords. Unlike exception handling in many languages—including Objective-C—error handling in Swift does not involve unwinding the call stack, a process that can be computationally expensive. As such, the performance characteristics of a throw statement are comparable to those of a return statement.
12:36:47	Araq	it installs mingw for you and works out of the box.
12:37:25	Araq	Alpha1220: that's just an implementation detail / performance characteristic. the import part here is that throws funcs do auto-throw when then call throws funcs
12:37:37	Araq	*when they call
12:37:51	bodie_	hm
12:38:14	bodie_	so I guess threading on windows depends on pthread?
12:39:02	Araq	Alpha1220: in other words, the stack unwinding exists and is implicit. like in any other exception handling model.
12:39:04	bodie_	I'm not comfy w/ using mingw for windows native. I realize this is not an issue for everyone
12:39:15	bodie_	how big of a pain is it to set up for msvc?
12:39:26	bodie_	or is it just a "just generate the C and DIY"
12:39:36	Araq	edit the nim.cfg; cc = vcc
12:39:37	bodie_	alternatively, maybe can it be done using a nimscript build?
12:39:41	Araq	that's it.
12:39:46	bodie_	that's nice
12:39:53	*	Senketsu_ quit (Quit: WeeChat 2.1)
12:39:59	bodie_	so it doesn't depend on pthread?
12:40:00	Alpha1220	BTW are the plans for making seqs and strings default to empty instead of nil in plans for v1? That is the number 1 reason for my code crashes.
12:40:14	FromGitter	<arnetheduck> funny enough, swift offers both ways of handling them as well, they have a `try?` operator to turn an exception into a nil value
12:40:51	*	simonsolnes quit (Ping timeout: 255 seconds)
12:42:01	*	Senketsu_ joined #nim
12:42:54	bodie_	one of my bigger gripes with Rust is how people have written crates that have hard dependencies on a gnu build environment
12:42:57	Alpha1220	I would seriously prefer rust like `?` and Result for error handling but well idk how well it would work with overloaded operators.. Anyone with experience in Rust who can shed some light?
12:43:42	bodie_	I like the ? macro
12:44:02	bodie_	I don't like the ergonomics of Result and Option very much
12:44:03	*	Senketsu_ quit (Client Quit)
12:44:19	bodie_	it's better than nothing, but at the end of the day I think Go's solution is simpler
12:44:54	bodie_	checking a value in a boolean expression is simpler than pattern matching, especially with inner values
12:45:16	bodie_	that doesn't mean result / option aren't _better_ but _simplicity_ is a big facet of usability
12:45:23	bodie_	personal opinion here
12:46:16	bodie_	I'd love it if Go had some kind of macro for eliding the boilerplate of error checking though
12:47:08	bodie_	maybe something like "try" without the runtime semantics of exceptions....
12:47:46	Araq	Alpha1220: yup, planned for v1.
12:50:04	*	Alpha1220 quit (Ping timeout: 260 seconds)
12:50:19	bodie_	do you know whether there's a nanogui binding for nim?
12:50:40	FromGitter	<mratsim> by the way @alehander42 what’s the rationale for using urandom instead of random?
12:51:55	*	xkapastel joined #nim
12:52:40	FromGitter	<mratsim> @Alpha1220, why would it be a problem for overloading?
12:53:31	FromGitter	<alehander42> @mratsim well, I think I just thought nim-random is a richer lib than the builtin random module
12:53:50	FromGitter	<mratsim> `.` being overloadable today is a an issue granted but I think ? being overloadable is fine, some libraries might return multiple error types and you can always dispatch on type
12:54:17	FromGitter	<mratsim> ah, so it’s nim-random that pulls the random number from /dev/urandom
12:54:20	FromGitter	<alehander42> but I guess `/dev/urandom` is more useful, as it always returns stuff (it doesn't block: the bug you encountered was just related to io bug in older version of the lib I forgot to update)
12:55:16	FromGitter	<alehander42> and I didn't really analyze e.g. how crypto-reliable would it be, as I just needed relatively random values for tests
12:55:43	FromGitter	<alehander42> as a whole, I intended to add some additional random sources and a way to tweak what distribution you want
12:55:54	FromGitter	<alehander42> but I haven't had a usecase for that yet
12:56:16	FromGitter	<mratsim> "Warning: Package 'quicktest' has an incorrect structure. The top level of the package source directory should contain at most one module, named 'quicktest.nim', but a file named 'new.nim' was found. This will be an error in the future.” just another day in nimble lol
12:56:17	FromGitter	<alehander42> (at least I think i've seen this kind of hooks in other libs)
12:56:38	FromGitter	<alehander42> ah this is real, I should remove this file :D
12:56:41	FromGitter	<mratsim> For me I just want the test to finish fast
12:56:46	Araq	for me their '?' macro smells like they really want a short notation to pass errors up the call stack. Which is what exceptions do with an even shorter notation.
12:57:06	FromGitter	<mratsim> ^
12:57:30	Araq	you can argue all day long that this single character makes software more reliable in the greater scheme of things.
12:57:37	FromGitter	<arnetheduck> Araq, `?` is visible, nothing is not
12:58:01	Araq	yeah here we go.
12:58:14	FromGitter	<narimiran> ? can be omitted, nothing can't :P
12:58:21	FromGitter	<arnetheduck> it's the fact that you actively have to put it there to appease the compiler that makes the difference
12:58:22	Araq	at some point the ? is just noise in the code
12:58:34	Araq	you write it down without thinking.
12:58:37	*	alpha1220 joined #nim
12:58:45	Araq	and that's exactly what you wanted to avoid in the first place.
12:59:08	bodie_	any code which can result in an error value should simply be a compile error :)
12:59:10	FromGitter	<arnetheduck> and that's where nim can do better - you can do a `check` block for example
12:59:33	FromGitter	<alehander42> what is `check` block?
12:59:33	Araq	you can't force "make me think about error handling" into programmers, they will do anything to avoid it.
12:59:44	shashlick	Araq: is https://github.com/nim-lang/Nim/pull/7677 ready to go?
12:59:47	FromGitter	<arnetheduck> so you already made this choice and argument with `discard` - how is this different?
12:59:51	FromGitter	<mratsim> Ah one thing, I’d like a fast_finish option so that it stops on first fails. Or give a summary of one example failing case and the number of failures. Otherwise it’s spamming the stdout.
13:00:13	alpha1220	Well, for me the biggest advantage of ? is mental peace. I know which functions can return error and I can put a question mark on each function call. I hate nested try-catch-finally blocks.
13:00:49	FromGitter	<arnetheduck> Araq, of course you can.. if people have to pay 10c for getting a plastic bag, they'll bring their own.. even if they earlier paid the same 10c indirectly
13:01:15	FromGitter	<mratsim> raises: [] might help @alpha
13:01:21	FromGitter	<mratsim> though it’s longer to write than ?
13:01:42	Araq	the granularity is much better, one .raises vs 1000s of ?s
13:01:51	*	alpha1220_ joined #nim
13:01:52	FromGitter	<krux02> Can someone tell the "Ron" in the comment section of this article, that references are nothing but hidden pointers and can as well be null. http://etodd.io/2017/03/29/thirteen-years-of-bad-game-code/
13:02:02	*	simonsolnes joined #nim
13:02:20	FromGitter	<krux02> I can't login
13:02:25	FromGitter	<arnetheduck> @alehander42 `check` block would be the equiv of `?`, but for a series of calls that return errors, so you don't have to litter the code
13:03:26	codevantage	krux *gasp* someone is wrong on the internet XD
13:03:27	alpha1220_	mratsim Araq, that is the point. If you have Result, you are forced to use ? to unwrap. But no-one is forced to write raises:[], I am sure not even the stdlib uses it consistently, neither do any nimble package. So saying that raises:[] can do it is actually useless if people rarely use it.
13:03:44	FromGitter	<mratsim> check block in the wild: https://github.com/status-im/nim-mpint/blob/master/tests/test_comparison.nim#L22
13:04:11	FromGitter	<mratsim> yes but you are still forced to handle the exception
13:04:18	FromGitter	<mratsim> while you can discard Result or Status code
13:04:39	*	alpha1220 quit (Ping timeout: 260 seconds)
13:05:04	FromGitter	<alehander42> @arnetheduck but if I have a serie of calls, it's very probable that some of them will not return errors
13:05:19	FromGitter	<mratsim> anyway, for numerical computing Result and Error code are a no-go, for stuff like IO, networking so much things can fail exceptions are not suitable. Plus you need to recover
13:05:22	FromGitter	<alehander42> it's a very specific syntax optimization
13:05:52	FromGitter	<alehander42> `f?` seems better, as it doesn't make you remember two different syntaxes for the same thing
13:06:03	Araq	alpha1220_: so let's fix the stdlib and see how to patch selected Nimble packages.
13:06:27	Araq	then we can say from experience that Nim's approach sucks and we should abandon it.
13:06:34	*	ieatnerds quit (Remote host closed the connection)
13:06:54	FromGitter	<krux02> @mratsim numerical computations have NaN, which is basically a monad around the error case
13:06:58	*	ieatnerds joined #nim
13:07:00	FromGitter	<arnetheduck> Araq, to know where to fix, the compiler has to tell you.. that's why we're having this discussions
13:07:02	FromGitter	<krux02> it just doesn't tell you the error
13:07:08	Araq	heck, I don't even like exceptions and yet I think it's too early to throw them away.
13:07:26	Araq	arnetheduck: the compiler DOES tell you.
13:07:36	FromGitter	<krux02> Araq: don't even think about throwing them away.
13:07:51	FromGitter	<krux02> but I would like to have the stack upside down
13:07:58	FromGitter	<mratsim> a queue?
13:08:02	FromGitter	<krux02> the way Nim puts the stack confuseses me
13:08:14	codevantage	why not have both?
13:08:34	FromGitter	<mratsim> “There is a macro for that”™
13:08:42	FromGitter	<arnetheduck> Araq, if the compiler told you (as a requirement) they would be there
13:08:55	FromGitter	<krux02> there is not a macro for everything
13:09:00	FromGitter	<krux02> and macros have disadvantages
13:09:04	*	ieatnerds quit (Remote host closed the connection)
13:09:07	FromGitter	<mratsim> it was a joke ;)
13:09:24	FromGitter	<krux02> for example I can't use the ast-pattern-matching macro to write the ast-pattern-matching marco
13:09:28	*	ieatnerds joined #nim
13:10:24	alpha1220_	Araq: On another note, why are object variants so clumsy to use and why is something like Patty not considered as part of the language?
13:10:28	codevantage	if exceptions could work in exactly the same way, then this would be just an implementation detail
13:11:13	Araq	alpha1220_: because I'm an idiot. next question.
13:11:23	Araq	:P
13:11:24	FromGitter	<alehander42> yeah, I also love the variant definitions of Patty :D
13:11:38	*	Snircle joined #nim
13:11:57	FromGitter	<alehander42> they also match very well with @krux02's pattern macthing
13:12:01	FromGitter	<krux02> Araq: You just insulted an important Member of the Nim community. Please excuse yourself.
13:12:09	alpha1220_	Come on, can we do this for 1.0? @Araq
13:12:24	alpha1220_	It will make nim a lot more pleasant than it is now
13:12:29	Arrrr	Wasn't patter matching for case objects planned before 1.0?
13:12:40	alpha1220_	and it goes together with the expressive and concise mantra of Nim
13:12:46	FromGitter	<krux02> alpha1220_: can you integrate Patty with ast_pattern_matching?
13:12:51	*	fredrik92 joined #nim
13:13:52	Araq	what you're describing here are the super negative aspects of an overly long horrible amateurish development process
13:14:07	alpha1220_	krux02, tbh I have never used macros, never needed them, so no I can't integrate them as of yet
13:14:31	Araq	so v1.0 is at the same time overdue and yet obviously not perfect.
13:14:53	alpha1220_	less amateurish than me. I never completed a project. Actually never did more than 1 month work on any project before abandoning it.
13:14:55	FromGitter	<krux02> alpha1220_: It is time that you do it. In my opinion the most important feature of Nim. You can't just dismiss it.
13:16:07	Araq	all we can do is either delay v1.0 by another 5 years or release what we have and pleasently surprise you by a v2.0 that rocks.
13:16:17	*	couven92 quit (Ping timeout: 248 seconds)
13:16:37	Araq	the successful software project all seem to do the latter.
13:16:38	FromGitter	<alehander42> @krux02 my long term hope was that Patty's variant etc macros and ast_pattern_matching more general `match` macros will be combined in the stdlib :D
13:17:03	codevantage	keep nim backwards compatible pls
13:17:28	alpha1220_	krux02: I would love to but you know where I live, nearly all jobs are on android/ios/web so even though I dislike all of them, yet I have to focus on them for the time being (as college time is gonna finish soon. So very little time to play around with Nim and the likes.
13:17:56	FromGitter	<krux02> @alehander42 I dismissed Patty pretty quickly, because I saw that it could not do ast pattern matching and was also not designed to do so.
13:18:06	FromGitter	<narimiran> if only all this energy were directed towards closing the bugs.... ;)
13:18:16	FromGitter	<ephja> will more cash end up in the hands of zah at some point? 😄
13:18:22	FromGitter	<mratsim> 50 bugs closed in the past week
13:18:38	FromGitter	<alehander42> @krux02 otherwise I think patty and ast_pattern.. 's matching dsl-s look relatively similar
13:18:39	FromGitter	<narimiran> @mratsim those were mostly duplicates, from what have i seen....
13:18:57	Araq	mratsim: closed, but "fixed" is another topic.
13:19:00	FromGitter	<mratsim> btw, is there a way to have nimble task only dependencies?
13:19:02	FromGitter	<narimiran> but nonetheless i was very pleasantly surprised to see the number go below 1300
13:19:16	FromGitter	<mratsim> I’d like to add quickcheck as a dependency but only for testing
13:19:22	FromGitter	<narimiran> yeah, should have said 'fixed' above
13:19:41	FromGitter	<alehander42> I think with a bit of work one could adapt the patty API to work well for the ast, but I just decided your lib has implemented more of the matching logic
13:19:44	alpha1220_	Araq: Well I have no problems waiting for v2 since it is just a number but well if you change it too much either you are gonna break everything or nim is gonna be the big mess that is c++. On the other hand, languages like Rust changed rapidly before 1.0 and no one complained (Yeah it had mozilla and that is one of the reasons but still.)
13:20:22	FromGitter	<mratsim> no one complains, not sure, it’s a mess with the #[cfg(features)] flag right now
13:20:30	FromGitter	<mratsim> or whatever it is
13:20:31	alpha1220_	BTW any plans for forming an organization around Nim and raise financial support so that there can be more official full time developers?
13:20:39	FromGitter	<mratsim> lots of old library only compile for Rust nightly
13:21:20	alpha1220_	The only complaints are regarding the over usage of nightly. But that is mostly for newer features. Breakages are pretty much uncommon in Rust atm.
13:22:16	FromGitter	<krux02> @alehander42 surely patty and ast_pattern_matching looks similar, structurally they are very different.
13:22:36	FromGitter	<krux02> Patty can only match very simple patterns
13:23:28	FromGitter	<alehander42> @krux02 exactly, I am talking about its "planned" API: (things that do not work yet)
13:23:56	FromGitter	<krux02> and I suspect that not including complexer patterns already made that library not forward compatible (there need to be breaking changes to generalize it)
13:24:29	Araq	alpha1220_: plans for this are ongoing, yes.
13:24:44	Araq	in fact, see you later.
13:25:39	alpha1220_	Oh great!
13:26:44	FromGitter	<alehander42> @krux02 one "future" example in patty I like is custom matchers: ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5adf30941130fe3d360f4403]
13:27:39	FromGitter	<alehander42> hm actually that's not a custom mactcher
13:28:40	FromGitter	<krux02> that is just impossible to support in that notation.
13:29:06	FromGitter	<krux02> Patty is not a general pattern matching library
13:29:10	FromGitter	<alehander42> anyway, still one could imagine similar custom matchers, e.g. ⏎ ⏎ macro myWeirdMatch(..): ⏎ # generates a condition and error handler ⏎ ... [https://gitter.im/nim-lang/Nim?at=5adf31265d7286b43a616a3e]
13:29:16	FromGitter	<alehander42> oh man wtf gitter
13:29:26	FromGitter	<krux02> Patty is a single level object variant matcher
13:30:06	FromGitter	<alehander42> I am talking about the original idea behind patty(which seems not implemented) https://github.com/andreaferretti/patty#things-that-do-not-work-yet
13:30:34	FromGitter	<alehander42> otherwise yeah, the syntax in the example is not so important, ignore this example
13:30:48	FromGitter	<krux02> If it would claim to be exactly that, I would not have any complains, but it sounds like it is pattern matching, but it is not, it is just a specific use case of it.
13:31:39	FromGitter	<alehander42> actually it even had the some plan for guards as yours `Circle(x: x, y: y, r: r) if r < 0:`
13:31:45	FromGitter	<krux02> Scala has a very neat way to do custom matchers
13:32:17	FromGitter	<alehander42> as I said, if it had all features implemented, it would be reasonably patter matcher
13:32:24	FromGitter	<alehander42> oh, what does scala do?
13:33:01	codevantage	does nim have a roadmap?
13:33:33	FromGitter	<krux02> scala has ``def unapply(object: S): Option[T]``
13:33:49	FromGitter	<narimiran> codevantage: making a roadmap is on the roadmap :P
13:33:52	FromGitter	<krux02> when you implement that for your type T, then this function will be called
13:34:22	FromGitter	<mratsim> there used to be a TODO.md
13:34:44	FromGitter	<krux02> it should return either the matched object or `None`
13:34:46	FromGitter	<mratsim> it has been changed to RFCs. But there is still a Things for V1 document
13:34:59	*	alpha1220_ quit (Ping timeout: 260 seconds)
13:35:06	FromGitter	<krux02> and it get's an object, which is basically RootObj
13:35:25	*	codevantage quit (Read error: Connection reset by peer)
13:35:56	*	codevantage joined #nim
13:36:48	FromGitter	<krux02> wait, my memory is weakening
13:36:54	FromGitter	<krux02> here is an article: http://danielwestheide.com/blog/2012/11/21/the-neophytes-guide-to-scala-part-1-extractors.html
13:38:48	codevantage	so no more todo ~
13:39:35	FromGitter	<krux02> codevantage: nim as bugs
13:39:43	FromGitter	<krux02> just check the issues list.
13:39:56	FromGitter	<krux02> codevantage: I think one item on the TODO list is concepts.
13:40:00	FromGitter	<krux02> I don't like them
13:40:43	codevantage	i think concepts works well with nim
13:40:56	FromGitter	<krux02> I disagree
13:41:00	codevantage	especially with the ufc
13:41:06	FromGitter	<krux02> ufc?
13:41:12	FromGitter	<ephja> what else?
13:41:15	codevantage	unified fuction call
13:41:31	FromGitter	<krux02> that is great, yes.
13:41:40	FromGitter	<krux02> but for my code I am much better off not using concepts
13:41:40	codevantage	maybe im getting the acronym wrong
13:41:59	FromGitter	<krux02> I also get compilation errors, the same was as if i would use concepts
13:42:06	FromGitter	<krux02> just that I don't need to deal with concepts.
13:43:07	codevantage	concepts are like contracts
13:43:37	codevantage	i feel
13:45:53	shashlick	has strutils.escape been deprecated?
13:48:27	codevantage	concepts are nice to work with already existing code
13:50:50	FromGitter	<krux02> codevantage: don't do too much with it. They are experimental and it is very likely that the syntax changes.
13:51:07	FromGitter	<krux02> shashlick: just look it up
13:51:46	shashlick	Found the PR, there's no replacement yet though
13:52:06	FromGitter	<ephja> wasn't it possible to have multiple statements in a => lambda
13:52:11	FromGitter	<krux02> it is deprecated
13:53:36	codevantage	krux i just use them as a way to ensure that only certain types interact with my procs. if i wanted to do it another way id have to spend a lot more time
13:55:17	FromGitter	<alehander42> @krux02 *very likely*
13:55:26	codevantage	beside i just wanted to experiment with it, i think its a good match for the problem
13:57:45	FromGitter	<krux02> @alehander42 The current concept implementation confuses type expressions with value expressions.
13:58:07	FromGitter	<krux02> to disambiguate them, the syntax has to be changed in a breaking way.
13:58:25	*	SenasOzys quit (Ping timeout: 248 seconds)
14:00:59	FromGitter	<krux02> the concept declaration "dynamically" decides if something is meant to be a type or a value.
14:01:06	*	nsf quit (Quit: WeeChat 2.0.1)
14:01:34	*	leorize quit (Ping timeout: 265 seconds)
14:03:37	codevantage	oh not goof
14:04:07	codevantage	then ill have to add checks or use patterns
14:05:28	*	SenasOzys joined #nim
14:07:15	*	athenot_ quit (Remote host closed the connection)
14:07:35	*	athenot joined #nim
14:09:44	FromGitter	<krux02> attempt to call undeclared routine 'inline'
14:10:01	FromGitter	<krux02> am I now supposed to delete all my inline?
14:13:41	*	leorize joined #nim
14:17:45	FromGitter	<alehander42> the concept syntax is a long topic , I don't have time to discuss it, but it works well for me
14:19:00	*	ftsf joined #nim
14:20:12	ftsf	dom96, choosenim release binary on windows detected as having a trojan "Trojan:Win32/Tiggre!plock" probably a false positive but might be worth looking into?
14:21:21	*	rockcavera quit (Ping timeout: 248 seconds)
14:27:12	FromGitter	<mratsim> @alehander42 congrats quickcheck helped me detect aliasing issues that only occurs in release mode: https://github.com/status-im/nim-mpint/issues/23
14:33:32	FromGitter	<alehander42> @mratsim awesome, I'll try to fix the uint arg thing and the output until tomorrow when I have more time
14:36:44	*	endragor_ quit (Remote host closed the connection)
14:40:46	FromGitter	<mratsim> actually it’s Nim that has a bug? or C
14:49:25	FromGitter	<mratsim> it’s the -O3 optimization of Clang that was generating the issue
14:49:37	FromGitter	<mratsim> test case: ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5adf44002dfcfffd2b2cb164]
14:50:42	FromGitter	<mratsim> -O2 gives the same too by the way
14:51:11	FromGitter	<diegogub> Is it possible to go from string to enum?
14:52:10	FromGitter	<alehander42> weird, gcc does this for me too
15:01:36	*	CodeVantage_ joined #nim
15:03:37	*	user1101 quit (Quit: user1101)
15:04:17	*	miran joined #nim
15:05:36	*	cspar joined #nim
15:06:42	*	codevantage quit (Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org)
15:06:53	*	CodeVantage_ quit (Quit: Leaving)
15:07:07	*	CodeVantage joined #nim
15:10:40	FromGitter	<krux02> @diegogub: converting an enum to string is easy, so it is pretty easy to create a Table to do the inverse.
15:11:04	FromGitter	<diegogub> @krux02 ow, good idea..thanks
15:15:32	ftsf	isn't there already parseEnum?
15:16:11	ftsf	diegogub: https://nim-lang.org/docs/strutils.html#parseEnum,string
15:18:58	*	leorize quit (Ping timeout: 248 seconds)
15:20:18	FromGitter	<krux02> parseEnum does not use a table
15:20:26	FromGitter	<krux02> but yes you are right I did not know it
15:24:58	*	cspar quit (Ping timeout: 264 seconds)
15:28:59	FromGitter	<diegogub> @krux02 @ftsf I was trying to use it parseEnum..but it's not working..as I would expect
15:29:20	FromGitter	<diegogub> I have this enum: ⏎ """ ⏎ type FilterType = enum ⏎ CountryFilter = "Country" ⏎ DateFilter = "Date" ... [https://gitter.im/nim-lang/Nim?at=5adf4d4f2d0e228d7ba90536]
15:29:51	FromGitter	<diegogub> and doing parseEnumFilterType ("Country")
15:30:13	FromGitter	<krux02> do ``` around your code
15:30:14	FromGitter	<mratsim> that’s not an enum
15:30:22	ftsf	parseEnum[FilterType](input)
15:30:23	FromGitter	<mratsim> or can enum have string value?
15:30:39	FromGitter	<krux02> yes enum can have string values
15:30:44	*	cspar joined #nim
15:30:45	FromGitter	<mratsim> oh
15:30:54	ftsf	https://nim-lang.org/docs/manual.html#types-enumeration-types
15:30:57	FromGitter	<krux02> I wouldn't use it though
15:30:59	FromGitter	<mratsim> but $fooValue gives what in that case?
15:31:17	FromGitter	<diegogub> "Country" ⏎ @mratsim
15:31:17	shashlick	see how I've used parseEnum() here: https://github.com/genotrance/snip/blob/master/src/snip/keymap.nim
15:32:43	FromGitter	<krux02> I just found a bug
15:32:53	FromGitter	<diegogub> Im getting ValueError when using parseEnum with "Country" string
15:33:00	FromGitter	<diegogub> Exception
15:33:35	shashlick	parseEnum[FilterType]("CountryFilter") = FilterType.CountryFilter
15:34:07	shashlick	you can get the enum from its name, you cannot get it from its value I don't think
15:34:35	FromGitter	<diegogub> I see, what I'm using is a case like :
15:35:03	FromGitter	<diegogub> ```case "Country": ⏎ of $CountryFilter: ⏎ return CountryFilter ⏎ ``` [https://gitter.im/nim-lang/Nim?at=5adf4ea7109bb04332f7569c]
15:36:17	FromGitter	<krux02> I just need to look at strutils and I see bugs everywhere
15:36:33	FromGitter	<krux02> use the strutils package to sprinkle bugs in your codebase
15:36:48	*	simonsolnes quit (Ping timeout: 255 seconds)
15:38:24	FromGitter	<diegogub> @krux02 I'm using it a lot, let know if I can help you
15:38:42	FromGitter	<krux02> well, write your own version of it that does not have bugs in it.
15:38:50	FromGitter	<krux02> it's a horrible mess.
15:39:40	*	athenot quit (Read error: Connection reset by peer)
15:39:42	FromGitter	<krux02> strutils.escape should have been called "putRandomReversableEscapeCharacters"
15:39:50	*	athenot_ joined #nim
15:40:13	FromGitter	<diegogub> haha would be a very good crash course to nim
15:40:24	FromGitter	<krux02> strutils.cmpIgnoreStyle was designed to match nim.s style insensitive identifier matching
15:41:44	FromGitter	<krux02> but because that equality behavior changed and cmpIgnoreStyle did not adopt that new behavior because beckwards compatibility, cmpIgnoreStyle is now just useful for nothing and all code that is written for identifier matiching like in nim is wrong
15:43:55	*	Trustable joined #nim
15:44:39	*	ieatnerds quit (Quit: Leaving)
15:47:23	FromGitter	<krux02> https://github.com/nim-lang/Nim/issues/7686
15:48:21	*	PMunch quit (Quit: Leaving)
15:55:40	dom96	ftsf: choosenim embeds a binary as static data so that's probably why. Many antiviruses consider that suspicious.
15:57:27	shashlick	how's it going dom96, been a while
16:00:12	*	CodeVantage quit (Read error: Connection reset by peer)
16:02:21	shashlick	did someone break the build? my PR is suddenly failing on a line I didn't change
16:02:26	*	Arrrr quit (Quit: Leaving.)
16:03:22	*	athenot joined #nim
16:04:43	*	simonsolnes joined #nim
16:06:30	*	athenot_ quit (Ping timeout: 255 seconds)
16:09:47	*	CodeVantage joined #nim
16:09:57	*	CodeVantage quit (Remote host closed the connection)
16:10:11	*	CodeVantage joined #nim
16:11:50	FromGitter	<krux02> shashlick is it a timeout?
16:12:56	shashlick	stack trace: (most recent call last)
16:12:58	shashlick	commands.nim(67)
16:13:00	shashlick	lib\pure\strutils.nim(2419) %
16:13:02	shashlick	lib\pure\strutils.nim(2362) addf
16:13:04	shashlick	lib\pure\strutils.nim(2304) invalidFormatString
16:13:05	*	CodeVantage quit (Client Quit)
16:13:06	shashlick	lib\pure\strutils.nim(2304, 3) Error: unhandled exception: invalid format string
16:13:08	shashlick	FAILURE
16:13:10	shashlick	Command exited with code 1
16:13:17	*	CodeVantage joined #nim
16:16:01	FromGitter	<krux02> @diegogub: I told you strutils if full of bugs, do you see it?
16:17:13	*	NimBot joined #nim
16:17:37	FromGitter	<krux02> it is in strutils
16:17:39	FromGitter	<krux02> isn't it?
16:19:27	FromGitter	<data-man> @Araq @dom96 ⏎ In newer versions of SQLite was added so many new useful features. ⏎ If I add to updated wrapper something like ```when defined(useSQLiteNewApi)``` or ```when defined(useSQLiteApiVersionXXX)``` it will be ok? ⏎ A new API is needed for some of my new PRs. [https://gitter.im/nim-lang/Nim?at=5adf590f6d7e07082b2c0063]
16:22:59	*	ftsf quit (Quit: Leaving)
16:24:24	FromGitter	<mratsim> in before all Nim packages must be compiled with -d:useSQLiteNewApi —newruntime -d:fooSomething -d:whatAboutMe
16:27:05	*	CodeVantage quit (Ping timeout: 240 seconds)
16:29:20	FromGitter	<data-man> For Windows It would be preferable if SQLite will be updated in dlls.zip
16:39:33	*	natrys joined #nim
16:42:20	*	riidom joined #nim
16:45:24	*	codevantage joined #nim
17:01:24	*	athenot quit (Ping timeout: 255 seconds)
17:14:15	*	athenot joined #nim
17:21:23	*	endragor joined #nim
17:25:53	*	endragor quit (Ping timeout: 248 seconds)
17:29:06	Araq	data-man: fine with me
17:30:15	FromGitter	<data-man> Which variant? With NewApi or ApiVersion?
17:33:40	Araq	when not defined(onlyOldSqliteApi)
17:41:27	*	jrbrt joined #nim
17:45:46	*	SenasOzys quit (Remote host closed the connection)
17:47:23	*	Trustable quit (Quit: Leaving)
17:54:19	*	jrbrt quit (Quit: jrbrt)
17:55:06	FromGitter	<data-man> But ```onlyOldSqliteApi``` not defined by default. Need to add sqlite3.nim.cfg with this define?
17:55:57	*	jrbrt joined #nim
17:57:06	*	jrbrt quit (Client Quit)
17:58:08	*	nsf joined #nim
17:58:18	*	SenasOzys joined #nim
17:59:01	FromDiscord	<treeform> It looks like seq.add gets slower and slower as you add items too it. Where is this implemented? Does not appear that it is amortized? Is there an amortized version of seq.add?
17:59:12	FromDiscord	<treeform> Is this it? https://github.com/nim-lang/Nim/blob/8b0ea4b501125880bce54f20839b443973bba0c4/lib/core/seqs.nim#L70
18:00:07	FromDiscord	<treeform> This has the funny side effect of making JavaScript version x1000 faster 😃 then C version.
18:03:45	Araq	treeform: it is amortized.
18:05:47	FromDiscord	<treeform> Where?
18:06:04	FromGitter	<data-man> @treeform: If you know the approximate number of items, then a preallocating will be faster.
18:06:09	FromDiscord	<treeform> I would expected it to be done on this line:
18:06:11	FromDiscord	<treeform> https://github.com/nim-lang/Nim/blob/8b0ea4b501125880bce54f20839b443973bba0c4/lib/core/seqs.nim#L66
18:06:16	FromDiscord	<treeform> some thing like
18:06:28	FromDiscord	<treeform> if x.len >= x.cap: resize(x.len*2)
18:06:57	FromDiscord	<treeform> oh maybe its this line: https://github.com/nim-lang/Nim/blob/8b0ea4b501125880bce54f20839b443973bba0c4/lib/core/seqs.nim#L85
18:07:20	FromDiscord	<treeform> data-man, yeah I don't know the size ahead of time.
18:09:45	FromGitter	<krux02> you all probably don't care about it at all, but I now have nice error messages like this generatey by my pattern mactichng library: http://ix.io/18Dm
18:11:22	Araq	https://github.com/nim-lang/Nim/blob/devel/lib/system/sysstr.nim#L249
18:13:27	Araq	krux02: I PM'ed you on gitter
18:15:35	FromGitter	<krux02> Araq: I PM'ed you back on gitter
18:18:00	FromDiscord	<treeform> it appears to be much faster to add to small seq, then its really slow to add to large ones
18:18:19	FromDiscord	<treeform> I could use seq of seq to speed it up maybe... like a rope
18:20:05	Araq	treeform: check the growth pattern please
18:20:15	FromGitter	<data-man> @Araq: In the current sqlite's wrapper functions look like this: ```proc sqlite3_xxxx(para1: someType, para2: someType, ...): someType``` ⏎ But c2nim starts the name of the parameter with 'a' and starts numbering for some reason from 2. ⏎ Maybe add a ```--paramsPrefix:PREFIX``` parameter to c2nim?
18:20:25	Araq	we can check V8 to see why it's much faster in JS land
18:20:45	*	natrys quit (Ping timeout: 264 seconds)
18:21:05	FromDiscord	<treeform> Araq, I am not sure what you want me to do?
18:21:31	FromDiscord	<treeform> I don
18:21:57	FromDiscord	<treeform> I don't understand how the code that you linked amortizes it.
18:22:15	Araq	treeform: make 'resize' output its values, perhaps
18:22:33	Araq	well it growths the seq by the factor of 1.5
18:26:07	*	jrbrt joined #nim
18:26:45	Araq	data-man: sure but that's not important
18:31:31	FromDiscord	<treeform> Araq, I am probably not good enough at the optimizations. I think its some sort of cache locality thing. It amortizes fine but still slows down. I think I just ran the JavaScript thing wrong so it proved to be much faster, but it probably did not do the work it should have done.
18:31:35	*	cspar quit (Ping timeout: 240 seconds)
18:32:24	*	natrys joined #nim
18:32:42	FromDiscord	<mraxilus> does nim have go-like interfaces? those were great
18:33:02	FromGitter	<data-man> @Araq: For visual compatibility :)
18:36:36	Araq	mraxilus: "interfaced" Nimble package for now
18:41:20	*	smt` quit (Remote host closed the connection)
19:00:14	*	nsf quit (Quit: WeeChat 2.0.1)
19:14:39	*	natrys quit (Ping timeout: 260 seconds)
19:14:57	*	Jesin joined #nim
19:15:07	*	jrbrt quit (Quit: jrbrt)
19:16:05	*	Vladar quit (Quit: Leaving)
19:16:56	*	jrbrt joined #nim
19:17:00	*	jrbrt quit (Client Quit)
19:25:22	*	jjido joined #nim
19:25:42	*	jrbrt joined #nim
19:28:17	*	athenot quit (Read error: Connection reset by peer)
19:31:21	*	athenot joined #nim
19:37:36	*	endragor joined #nim
19:42:22	*	endragor quit (Ping timeout: 264 seconds)
19:48:22	*	Jesin quit (Quit: Leaving)
19:52:43	*	noonien quit (Quit: Connection closed for inactivity)
19:52:49	*	Sembei quit (Quit: WeeChat 2.2-dev)
19:55:42	*	Sembei joined #nim
19:56:27	*	natrys joined #nim
19:56:51	*	CodeVantage_ joined #nim
19:58:10	*	Sembei quit (Client Quit)
19:59:00	*	jrbrt quit (Quit: jrbrt)
20:02:28	*	rockcavera joined #nim
20:16:21	*	yglukhov quit (Remote host closed the connection)
20:17:16	*	Sembei joined #nim
20:17:54	*	rokups quit (Quit: Connection closed for inactivity)
20:18:08	*	Sembei quit (Client Quit)
20:20:17	*	Sembei joined #nim
20:32:50	*	jjido quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
20:33:24	*	Sembei quit (Quit: WeeChat 2.2-dev)
20:34:07	*	Sembei joined #nim
20:36:49	FromGitter	<data-man> @dom96: What do you think about #7549 ? Once Araq wrote that: ⏎ ⏎ > well it was a -1 from me, dom96 and data-man gave a +1 and so it's accepted ⏎ ⏎ So, if you like this idea, your "+1" will be decisive. Probably. :) ... [https://gitter.im/nim-lang/Nim?at=5adf95615f188ccc159d558d]
20:38:40	*	Sembei quit (Client Quit)
20:40:51	*	jrbrt joined #nim
20:42:42	*	Sembei joined #nim
20:47:43	*	Sembei quit (Quit: WeeChat 2.2-dev)
20:48:41	*	aziz joined #nim
20:50:34	*	CodeVantage_ quit (Quit: Leaving)
20:53:21	*	nsf joined #nim
20:53:22	*	miran quit (Ping timeout: 268 seconds)
20:54:42	*	aziz quit (Quit: Ex-Chat)
20:55:24	Araq	ah but that is different. I'll use my BDFL superpowers to prevent this from happening.
20:55:27	Araq	:P
20:55:50	*	Sembei joined #nim
20:56:00	Araq	we only add things to system that are useful.
20:56:20	Araq	and cannot be done elsewhere.
20:56:43	Araq	yes it's full of cruft because this policy is rather new. too bad for new features then.
20:57:08	Araq	the mistakes of the past don't justify new mistakes.
20:58:22	FromGitter	<data-man> Maybe add this to the algorithm module, at least?
20:59:15	FromGitter	<data-man> algosugar? :)
20:59:32	Araq	more like people_who_dont_understand_proc_pointer_syntax.nim ?
21:02:52	Araq	why not put your effort into making '->' part of system? (int, int) -> bool # binary integer predicate
21:04:13	FromGitter	<data-man> If we shall arrange voting, democracy will win? Or araqcracy will win? :-D
21:04:37	Araq	that depends on the topic.
21:05:14	Araq	we have a democracy here a la Ankh Morpork.
21:05:56	Araq	people get to vote and the most votes win. The only person who is allowed to vote is Araq.
21:07:40	FromGitter	<data-man> I did not read the 'Discworld' :(
21:09:16	Araq	well there are also movies and computer games about it. but the books are better. :-)
21:10:20	FromGitter	<alehander42> I watched a movie for one of the books
21:10:31	FromGitter	<alehander42> and it was very true to the original
21:10:35	FromGitter	<alehander42> down to dialog etc
21:10:47	FromGitter	<alehander42> and I hated it bec ause I already read the book, and it was just the same
21:11:08	FromGitter	<alehander42> so I guess that's a recomendation if you don't have time for any of the books :D
21:11:50	*	yglukhov joined #nim
21:13:16	FromGitter	<data-man> Other ideas for RFCs: ⏎ ⏎ 1) more flexible API for DB ⏎ 2) add API for the most popular key-value DBs ⏎ 3) multiple TLS/SSL backends for net (like cURL) [https://gitter.im/nim-lang/Nim?at=5adf9dec5d7286b43a63d989]
21:16:03	Araq	there are all Nimble package ideas, not nim RFCs.
21:16:57	FromGitter	<data-man> But net and db in the stdlib.
21:19:41	*	yglukhov quit (Remote host closed the connection)
21:19:46	Araq	and they suck
21:19:57	Araq	and need to be moved into nimble packages
21:20:17	*	yglukhov joined #nim
21:20:58	*	nsf quit (Quit: WeeChat 2.0.1)
21:22:47	FromGitter	<data-man> New re with PCRE2?
21:22:57	*	CodeVantage_ joined #nim
21:24:39	*	yglukhov quit (Ping timeout: 255 seconds)
21:32:05	*	xet7 quit (Read error: Connection timed out)
21:32:35	*	xet7 joined #nim
21:32:40	*	CodeVantage_ quit (Quit: Leaving)
21:35:12	*	athenot quit (Remote host closed the connection)
21:36:01	*	athenot joined #nim
21:44:08	*	BitPuffin quit (Remote host closed the connection)
21:46:27	*	vlad1777d joined #nim
21:48:57	*	smt joined #nim
21:49:56	codevantage	araq you a diskworld fan? :D
21:50:47	*	xet7 quit (Read error: Connection timed out)
21:51:35	*	xet7 joined #nim
21:52:46	Araq	yes.
21:53:20	*	simonsolnes quit (Ping timeout: 276 seconds)
21:55:22	FromGitter	<data-man> DisCworld :-D
21:59:12	Araq	Nimworld
21:59:40	codevantage	why there not such a web sitw
22:01:17	FromGitter	<data-man> I'm fan of 'The Wheel of Time'. Much cooler than 'The Lord of the Rings'. :-D
22:06:53	Araq	The Wheel of Time was too hard for me. that creepy fog that you have to fight constantly
22:07:07	FromGitter	<data-man> Thanks to Brandon Sanderson that he continued the great work of Robert Jordan.
22:07:18	Araq	graphics were good for its time though
22:08:03	codevantage	wheel of time? fog? what fog?
22:08:14	codevantage	isnt that another story?
22:09:14	FromGitter	<data-man> And I'm fan of Dan Simmons. :)
22:09:26	codevantage	so much classics
22:12:12	Araq	codevantage: wheel of time was also a game.
22:13:40	*	cspar joined #nim
22:14:13	codevantage	oh
22:14:30	codevantage	i thought it never happened
22:14:42	codevantage	iv never seen it here
22:14:50	codevantage	:(
22:22:55	*	jrbrt quit (Quit: jrbrt)
22:23:53	*	jrbrt joined #nim
22:25:14	codevantage	i think for v1 nim has to have a installer.
22:25:30	codevantage	i know we already got a system for that
22:25:58	codevantage	but id like to be able to download an installer from the website
22:26:29	*	CodeVantage_ joined #nim
22:27:46	*	codevantage quit (Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org)
22:29:00	Araq	CodeVantage_: what's wrong with unzip + finish or choosenim?
22:30:01	*	Sentreen quit (Quit: WeeChat 1.6)
22:30:50	CodeVantage_	It's not one click install
22:30:56	CodeVantage_	not noob friendly
22:31:08	Araq	I beg to differ.
22:31:11	CodeVantage_	A small wizard
22:31:22	CodeVantage_	it could set up docs
22:31:26	Araq	1. there is no such thing as a "one click install"
22:31:42	CodeVantage_	THere could be.
22:31:44	Araq	2. finish.exe is a wizard, you need to double click it like any other installer.exe
22:31:49	CodeVantage_	At least some walkthrough
22:32:04	Araq	did you run finish.exe?
22:32:09	CodeVantage_	Yes. its a wizard. I was thinking something more graphical obv
22:32:15	CodeVantage_	Yes Yes
22:32:23	FromGitter	<data-man> @CodeVantage: Are you a Windows user? :)
22:32:25	CodeVantage_	I even ran koch boot
22:32:26	Araq	what's wrong with it.
22:32:32	CodeVantage_	I is a windoz user
22:32:49	Araq	'koch boot' is not in our installation instructions.
22:32:59	CodeVantage_	Nothing, but I think an installer would be nicer.
22:33:04	Araq	follow the instructions please. and then come back and complain it's too hard.
22:33:14	CodeVantage_	It is in the install file
22:33:28	CodeVantage_	I did everything XD
22:33:53	Araq	seems more a documentation problem than a tool problem.
22:34:06	Araq	unzip, double click on finish.exe
22:34:13	Araq	that's the installation on Windows.
22:34:23	CodeVantage_	Installation on Linux/UNIX
22:34:24	CodeVantage_	...
22:34:24	CodeVantage_	Then run the following command::
22:34:24	CodeVantage_	sh build.sh
22:34:30	Araq	if you "did everything", then you did far too many unrequired steps
22:34:38	CodeVantage_	I read install file
22:34:54	*	yglukhov joined #nim
22:35:04	CodeVantage_	I ran koch cause I use a global nimcache
22:35:06	CodeVantage_	And it broke
22:36:56	Araq	makes no sense
22:36:58	CodeVantage_	nvm I read the instructions
22:37:01	CodeVantage_	Everything seems find
22:37:03	CodeVantage_	fine
22:37:25	Araq	we can give finish.exe a UI
22:37:36	CodeVantage_	That's what I wanted
22:37:51	Araq	but if the instructions are to blame, it's futile work.
22:38:06	CodeVantage_	And maybe finish could add nimble bin folder to path... In fact I'll go do that now
22:38:21	CodeVantage_	I didn't read the right instructions
22:38:29	CodeVantage_	Not the instructions fault :|
22:40:33	*	yglukhov quit (Ping timeout: 248 seconds)
22:45:31	FromGitter	<data-man> What dlls will be needed for UI? For what platforms? With what version of msvcrt? etc. :-D
22:45:38	*	CodeVantage_ is now known as CodeVance
22:45:47	CodeVance	ISn't there already iup
22:45:48	FromGitter	<dom96> We'll get a UI
22:45:52	CodeVance	there's also nigui
22:46:01	CodeVance	which is platform specific
22:46:06	CodeVance	windows and linux
22:46:08	FromGitter	<dom96> If nobody creates one I will once I'm using windows more again
22:46:40	FromGitter	<data-man> Maybe based on TUI?
22:47:51	Araq	I wrote one based on my libui
22:48:23	Araq	but it never made it into any release because it's problematic
22:49:04	Araq	to integrate the download loop into the UI's event loop is fragile
22:49:31	Araq	and an ugly wizard that works is better than a sexy one that hangs.
22:49:43	FromGitter	<data-man> So terminal's based is more preferable.
22:50:11	*	smt quit (Read error: Connection reset by peer)
22:50:14	Araq	well it's lipstick on a pig.
22:50:28	Araq	"preferable" by whom? by me? sure.
22:50:38	Araq	by our users? doubtful.
22:50:56	*	smt joined #nim
22:51:18	FromGitter	<data-man> With Ribbon UI :-D
22:51:20	CodeVance	Have you seen nigui?
22:51:22	*	ieatnerds joined #nim
22:51:26	ieatnerds	Hey!
22:51:38	*	natrys quit (Quit: natrys)
22:51:41	FromGitter	<data-man> Hi!
22:52:29	ieatnerds	Just a few questions as per usual I guess. or my company we have an in house database that we would like to encrypt. I just dont know how I'd go about that in nim.
22:52:35	ieatnerds	for*
22:52:36	Araq	data-man: use the Win API directly for it, maybe.
22:53:20	Araq	might give us the start for our own libui without its annoying dependencies
22:55:19	FromGitter	<dom96> Araq: can you think of pathological behaviour that could be used to make Nim's collector unreliable, or should I suggest Nim to this person? https://www.reddit.com/r/programming/comments/8edblq/rust_at_chucklefish_pdf/dxuvdfb :)
22:56:15	FromGitter	<dom96> FWIW my plan is to use the winapi directly. I have an interesting project idea actually
22:56:40	Araq	well you have to avoid cycles like in Rust. and then Nim's GC is *better* for games because it's incremental.
22:57:10	Araq	ok, maybe not better, but different.
22:57:23	Araq	depends on the involved data structures.
22:58:13	shashlick	Why not just use nsis
22:58:30	CodeVance	Araq nigui already uses the win api directly
22:58:46	FromGitter	<data-man> @ieatnerds: You can try https://github.com/sqlcipher/sqlcipher
22:58:49	Araq	as I keep telling you, I don't dislike GCs for their speed or lack of realtime-ness
23:00:20	Araq	I dislike them because they tend to lure you into bad designs and because they make interop with other systems harder
23:01:52	Araq	shashlick: nsis sucks, that's why. been there, done that.
23:02:07	ieatnerds	Thanks data-man!
23:02:10	Araq	CodeVance: what? huh? where?
23:03:00	CodeVance	https://github.com/trustable-code/NiGui/blob/master/src/nigui/private/windows/platform_impl.nim
23:03:06	CodeVance	and gtk for linux
23:03:41	shashlick	No doubt, but just for the UI q&a, everything else can use choosenim + finish
23:04:07	CodeVance	tell me about choosenim
23:04:19	CodeVance	Cause I haven't used it yet
23:04:30	shashlick	It's good but does only 32 bit
23:04:59	CodeVance	nigui?
23:05:13	CodeVance	I use it
23:05:21	CodeVance	I'm 64
23:05:45	shashlick	No choosenim
23:06:10	Araq	you're the oldest Nim user then here.
23:06:10	FromGitter	<data-man> @ieatnerds: Note: you will need to wrap several functions from sqlcipher's header.
23:07:03	CodeVance	Araq, lol
23:08:01	FromGitter	<mratsim> Trivia, Arraymancer’s releases are named after Fantasy books: https://github.com/mratsim/Arraymancer/releases/tag/v0.2.0
23:08:06	ieatnerds	@data-man: I'm well aware. I'm not really worried about that. I've just never actually had to work on cryptographic things before. So I feel a bit lost on where to start.
23:08:21	FromGitter	<mratsim> Don’t do your own crypto
23:08:37	ieatnerds	I'm also aware of that.
23:09:08	CodeVance	1st rule of crypto
23:09:10	FromGitter	<mratsim> well it’s fine to implement something following specs, but don’t create your own crypto scheme.
23:09:34	FromGitter	<mratsim> encrypting databases is a mess by the way.
23:09:47	tersec	mratsim: I understand that as guidance in general, but usually it's provided to people writing in a language with a wider set of pre-existing crypto libraries.
23:10:03	FromGitter	<mratsim> I worked with an encrypted database while in a Swiss bank … it’s super painful to deal with.
23:10:08	tersec	And even following specs can be hard with crypto, because of timing sidechannels, etc.
23:10:19	*	cspar quit (Ping timeout: 260 seconds)
23:10:47	tersec	There's also some irony in that for a language the choice of stdlib crypto of which amounts to ... md5 and sha1.
23:10:52	FromGitter	<mratsim> Try to find an encrypted database library in C, and wrap that.
23:11:12	ieatnerds	I'm just trying to stop us from have plain text csv files with client info. My boss has a few things he wants that no already-out-there solution offers, which is why were doing an in house thing.
23:11:22	FromGitter	<mratsim> For crypto in any case, feel free to use nimcrypto: https://github.com/cheatfate/nimcrypto
23:11:31	tersec	Sure, that's an especially tricky use case (as is anything where "encrypted" data has to be semi-online/live).
23:11:48	FromGitter	<mratsim> We’ll have it security audited at Status.
23:11:58	FromGitter	<mratsim> and it will be supported long-term
23:12:21	ieatnerds	@mratsim: Nice!
23:12:25	tersec	IIRC that has sha2, sha3/keccak (which, sure, Ethereum, so), and some other more modern algos?
23:13:04	ieatnerds	@tersec: Like I've said, I just dont know where to start. Luckily it's on the back burner so I have time to work on it.
23:13:49	FromGitter	<mratsim> the focus is on Ethereum crypto graph, including keyfiles (pbkdf2).
23:14:08	*	cspar joined #nim
23:14:08	FromGitter	<mratsim> cryptography*
23:14:38	tersec	ieatnerds: I guess structurally the trickiest thing I find about crypto stuff is the same as security stuff -- failure can't just be unit/integration/etc tested for. Failure's silent and commonly caused by intelligent, malicious actors.
23:14:54	FromGitter	<data-man> @Araq: Is NiGui a good candidate for inclusion to the stdlib?
23:15:18	Araq	no, but finish.nim is allowed to have nimble package dependencies
23:16:58	tersec	mratsim: is there any movement towards at least maybe replacing (there's good reason to have e.g., some crypto hashes in a stdlib, I'd claim) to replace md5 and sha1?
23:17:23	tersec	Or desire/willingness on your/Araq/Status's/etc part
23:18:22	FromGitter	<mratsim> Once that is sorted: https://github.com/nim-lang/Nim/issues/7337
23:18:29	FromGitter	<data-man> It would be ideal to update wrapper for IUP and to write a convenient high-level API for it. :)
23:19:09	CodeVance	is there a zen of nim?
23:19:11	tersec	mratsim: looks promising, would like to see that
23:19:14	shashlick	nimgen it
23:19:49	FromGitter	<data-man> @shashlick: For what?
23:20:02	shashlick	So that updates are almost automatic
23:20:35	*	yglukhov joined #nim
23:21:05	FromGitter	<mratsim> @shashlick, the only issue I have when browsing your nimgen libraries is that I don’t see the proc and types that I will get :/
23:21:23	shashlick	That's next on my to-do
23:21:35	shashlick	There's too many fun projects in Nim land
23:21:42	shashlick	Now I feel like making an installer
23:22:40	shashlick	Spring cleaning so fixed some defects
23:22:57	FromGitter	<data-man> IUP's API is very big. gl, scintilla, lua, image tools, OLE, plotting, web, ... What to exclude?
23:22:58	Araq	mratsim: I thought this RFC will be superseeded by "we use this in production now, can't be bad for the stdlib to adopt it as the standard"
23:23:45	shashlick	Given its already been wrapped in the past, should be easier to follow that right?
23:24:24	ieatnerds	Thank you all for the help, Time to do some(a lot) of reading!
23:24:51	Araq	CodeVance: my zen of Nim is "simple core plus macros"
23:25:00	Araq	and eventually it will be true.
23:25:29	*	yglukhov quit (Ping timeout: 260 seconds)
23:25:44	Araq	the guiding principle of the future developments.
23:25:59	CodeVance	you will reach the state of nim
23:26:36	*	ieatnerds quit (Quit: Leaving)
23:28:08	Araq	fortunately what 'simple' means is not static. Even Go is a pretty big language by the standards of the 70ies.
23:31:49	*	CodeVantage joined #nim
23:31:55	CodeVantage	nim is simple but expressive
23:33:21	*	CodeVance quit (Ping timeout: 255 seconds)
23:34:03	CodeVantage	how do I get the nimble package path?
23:34:08	FromGitter	<mratsim> @Araq, yeah we use it to produce our research :D
23:38:46	Araq	why the sudden interest in IUP? it's dead
23:38:54	Araq	no OSX support.
23:41:00	CodeVantage	tcl wrapper?
23:41:01	CodeVantage	XO
23:41:13	FromGitter	<data-man> @Araq: Dead? Last commit at r4819 | scuri | 2018-04-24 22:11:32 +0300 :)
23:41:49	Araq	they promised OSX support for as long as I promised v1
23:42:56	*	vivus joined #nim
23:44:03	FromGitter	<data-man> But development is more active than libui.
23:45:49	FromGitter	<data-man> And libui isn't pure C lib.
23:46:53	Araq	that never bothered me.
23:47:05	dom96	shashlick: nsis is so archaic it barely supports unicode
23:47:07	Araq	targetting osx with pure C is kinda masochistic
23:48:08	dom96	Araq: I was just looking for an excuse to mention Nim again on Reddit, your reply doesn't really give that
23:48:47	FromGitter	<data-man> What then? Using opengl for all platforms?
23:49:13	Araq	we need to release an article for reddit/HN, not hijack other threads
23:51:03	dom96	Now that I've got a keyboard I can share my idea for an installer library for Windows: take any CLI app and convert it into a wizard.
23:51:15	dom96	This should in theory allow me to take choosenim and turn it into a wizard automagically.
23:51:41	dom96	Each stdin question would be converted to a GUI question in the wizard
23:51:42	CodeVantage	I think GNU already did something similar with man
23:52:08	dom96	Who knows how well that would work in practice, but it's a worthy project to pursue
23:52:32	CodeVantage	most CLI is simple enough
23:52:45	CodeVantage	bool range string enum
23:52:47	CodeVantage	done
23:52:49	dom96	Araq: yes, write some more articles, your last one did very well on HN (not sure if you're aware)
23:53:22	dom96	but maybe focus on things Nim already has, instead of what this mythical version which we do not speak of will bring
23:54:08	dom96	CodeVantage: yep, the library could scan the CLI apps `--help` and generate a nice checkbox filled screen automatically too
23:54:38	FromGitter	<data-man> @dom96: automagically? Every day I learn many new words. It's cool! :)
23:54:54	dom96	It should give a convincing GUI installer 80% of the time :)
23:55:03	dom96	data-man: hehe, you probably won't find that in a dictionary
23:55:25	CodeVantage	why even bother with reddit, why not ycombinator?
23:56:40	Araq	I think I should do it Trump style. win the election with bullshit twitter messages.
23:57:21	CodeVantage	nim needs to go viral
23:57:29	CodeVantage	but we need a cute animal for that
23:57:30	Araq	"Hey, Nim what's up, just replaced 10 stdlib iterators with a single for-loop macro"
23:57:35	dom96	You should get Twitter
23:57:52	Araq	bah
23:57:59	dom96	You can CC random Rust/Swift/D devs and tell them they're wrong :P
23:58:15	CodeVantage	this for nim https://xkcd.com/353
23:58:31	Araq	hmm I am unable to use twitter language
23:58:53	CodeVantage	Just put # before every keyword]
23:59:00	CodeVantage	you'll be fine
23:59:16	dom96	Araq: If Trump can do it... :P
23:59:30	CodeVantage	trump is a genius
23:59:56	FromGitter	<data-man> @dom96: Yeah, "Nimwarts School of Witchcraft and Wizardry" (already was) :D