<< 01-01-2019 >>
00:00:06*ghost64 quit (Quit: See you!)
00:01:12*ghost64 joined #nim
00:01:31shashlickI'm here
00:01:36shashlickStill 6pm here
00:04:35Araqhttps://github.com/nim-lang/Nim/blob/devel/compiler/scriptconfig.nim#L41 you need to do something like this
00:05:50shashlickI'm familiar with that file, but how do I restrict it to nimscript only and not compile time
00:06:39Araqregister this stuff in a helper proc that you call from main.nim
00:07:16Araqcontext: commandEval()
00:08:03shashlickOk I'll take a deeper look
00:08:37shashlickI presume you are talking about adding this to Nim proper and not custom fork or compiling in compiler
00:10:26*zachk quit (Changing host)
00:10:26*zachk joined #nim
00:12:06Araqyeah, however
00:12:53Araqnotice that exposing 'stdin' will be harder than specializing stdin.readLine as a single operation, proc readAllFromStdin or something
00:14:57Araqbut I guess you can 'cast' the 'File' to an int64 and use the intVal field internally
00:16:14shashlickSo adding to setupVM won't expose stuff to macros and other compile time stuff?
00:16:57AraqI don't remember
00:17:09AraqI think setupVM is what everything can use :-/
00:23:16shashlickOk I'll experiment and keep you posted
00:27:21FromGitter<zacharycarter> I'm eager to see whatever you come up with - not entirely sure what your goal(s) is but any VM stuff interests me
00:36:29Araqwhat do you wanna know? use a register based VM with "super instructions". best design for performance and comparable in complexity to the terrible "stack based" VMs
00:41:11*kapil____ quit (Quit: Connection closed for inactivity)
00:41:21FromGitter<zacharycarter> mm just the topic interests me
00:42:19FromGitter<zacharycarter> I've read through Nim's VM code a bit and extended it to support a few things
00:42:44FromGitter<zacharycarter> kind of like how nimble does - so that I could write my own syntax for config files - but this was a silly experiment
00:43:19FromGitter<zacharycarter> I didn't try to actually make it support any more of the stdlib, which would have been a better use of my time - but I'm sure much more difficult as well
00:51:19Araqnot really difficult, but tedious and in the end that BDFL asshole may reject it
00:56:56FromGitter<timotheecour> Woohoo! https://github.com/nim-lang/Nim/pull/10150 `enable FFI at CT` works !
01:03:23Araqwow, impressively clean patch :-)
01:03:59Araqtoo bad the stdlib is littered with 'var THIS_IS_A_CONST {.importc.}: cint' hacks
01:07:50Araqhmm go to sleep or try out my little alternative exception handling codegen strategy
01:17:04FromGitter<Clyybber> @timotheecour Congratulations, good job and a happy new year :D
01:25:31rayman22201very cool new years gift timotheecour!
01:25:57rayman22201happy new years to those who rotated into the new year already :-)
01:38:28FromGitter<zacharycarter> @timotheecour nice!
02:01:21shashlickBrilliant!
02:01:31shashlickHappy new year folks
02:04:38FromGitter<arnetheduck> 4 hours to go from where I'm looking, but likewise - HNY!
02:17:17FromGitter<timotheecour> thanks! happy new year ya'll! meilleurs voeux
02:26:35*zachk quit (Quit: Leaving)
02:40:25FromGitter<zacharycarter> didn't realize I could do this
02:40:31FromGitter<zacharycarter> but this is pretty sweet - ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5c2ad31fbabbc178b20b46d1]
03:00:34*banc quit (Quit: Bye)
03:16:21*banc joined #nim
03:19:15*glassofethanol joined #nim
03:19:24glassofethanolHappy new year!
03:19:34*glassofethanol left #nim (#nim)
03:22:26*xace quit (Read error: Connection reset by peer)
03:22:43*xace joined #nim
03:26:07*kapil____ joined #nim
03:28:55*vlad1777d joined #nim
03:55:31FromDiscord_<citycide> Araq, slashlick: > what's your github account again? genocide ?
03:55:31FromDiscord_<citycide>
03:55:31FromDiscord_<citycide> That sounds like a mashup of myself & geno :)
04:23:20*thumbs_scum quit (Ping timeout: 272 seconds)
04:28:03FromDiscord_<aolko> hey there
04:28:31FromDiscord_<aolko> i've been told nim is more `developer friendly` than rust
04:28:42FromDiscord_<aolko> so i've decided to check
04:28:59FromDiscord_<aolko> how bad is the windows tooling btw?
04:30:14FromDiscord_<aolko> also how low is the entry level?
04:32:01Calinouhappy new year!
04:32:32FromDiscord_<aolko> and yes hny 2019
04:34:49rayman22201aolko windows support is pretty good. The BFDL Araq uses windows as his primary machine and visual studio code is well supported.
04:35:14FromDiscord_<aolko> how about intellij plugins?
04:35:47shashlick@citycide: :)
04:36:20shashlick@aolko Nim works very well cross platform
04:36:43FromDiscord_<aolko> suprisingly, didn't found anything
04:36:47FromDiscord_<aolko> surprisingly, didn't found anything
04:37:01FromDiscord_<aolko> at least on the website
04:37:07rayman22201intellij is not popular in the community so I don't think there is good support for it.
04:37:10shashlickI've not heard of anyone using intellij
04:37:40FromDiscord_<aolko> but i work in jetbrains ides, vscode just doesn't have those sweet features
04:38:36rayman22201contributions welcome lol. There is active work on getting good language server protocol support.
04:39:01rayman22201then you could just use an intellij lsp plugin
04:39:16FromDiscord_<aolko> anywho, what's with nim's arrays, why they are so whacked?
04:39:25FromDiscord_<aolko> (compared to ruby and php)
04:39:40shashlickLooks like there was something earlier
04:39:42shashlickhttps://plugins.jetbrains.com/plugin/7829-nim-language-support
04:39:46rayman22201ruby and php are the "whacked" ones imo lol
04:39:52shashlickNo idea if it works though
04:40:18shashlickWhat is the issue with Nim arrays
04:40:21FromDiscord_<aolko> no, as a php webdev i can tell you, it's arrays are the best ones, they are also universal
04:40:36rayman22201what is your issue with Nim arrays?
04:40:52FromDiscord_<aolko> also, `last update: Sep 26, 2015`, ouch
04:41:25FromDiscord_<aolko> why splitting them in "tuples" and "sequences"?
04:41:45rayman22201neither of those are arrays. Those are different data structures.
04:42:25FromDiscord_<aolko> oh, makes it worse tho
04:42:34shashlickNote that Nim is a compiled language, not interpreted like ruby and php
04:42:50rayman22201Nim is closer to C++ than those languages. It is statically typed and compiled.
04:42:54FromDiscord_<aolko> crystal tends to disagree, while borrowing ruby's arrays
04:43:50rayman22201Nim outperforms Crystal. That "universality" has costs.
04:44:05FromDiscord_<aolko> having these arrays would be great - `["ones","that" => "can",["fit","multiple","types" => "yknow"]]`
04:44:22FromDiscord_<aolko> having these arrays would be great - `["ones","that" => "can",["fit","multiple","types" => "yknow"],123]`
04:44:53shashlickYou pay a heavy cost for duck typing
04:45:05rayman22201yup. That has major runtime costs
04:45:19FromDiscord_<aolko> i can only heavily shrug, besides sugar would be neat
04:45:50FromDiscord_<aolko> even python has braces, as a pip package
04:45:56rayman22201Nim can do similar things with variant types, which are checked at compile time and compile down to C enums.
04:46:03rayman22201Rust does not have this
04:46:12FromDiscord_<aolko> see: https://pypi.org/project/brackets/
04:46:57FromDiscord_<aolko> it would be fitting that this would be a part of `sugar` package
04:47:11shashlickAnything is possible with Nim macros
04:47:11rayman22201It's way too expensive. We don't want.
04:47:22rayman22201You can totally make a macro to do it though
04:47:24shashlickSomeone should be motivated to build it
04:47:41shashlickI do miss the super flexible python dictionaries
04:47:43FromDiscord_<aolko> you may not want that, understandably, unfamiliar people might
04:47:50FromDiscord_<aolko> and newbies too
04:48:14FromDiscord_<aolko> i have no clue on haw to make such macros...or any nim macros
04:48:15shashlickBut I really like the performance
04:48:18FromDiscord_<aolko> how*
04:48:20rayman22201newbies should learn the costs of their "sugar". not just take it for granted. but that's just my opinion
04:48:27shashlickThat's why I ditched python
04:48:53FromDiscord_<aolko> (meanwhile delphi is very newbie friendly and very performant)
04:49:07rayman22201lol. Delphi is not performant
04:49:10shashlickPlus the fact that so many of my bugs get detected at compile time
04:49:18FromDiscord_<aolko> it is, worked with it
04:49:24rayman22201so did I
04:49:26shashlickNim does have some Delphi legacy
04:49:54rayman22201Nim was originally written in Pascal, so we know all about that style of language.
04:49:56shashlickhttps://github.com/nim-lang/pas2nim
04:50:09FromDiscord_<aolko> can it do VCL?
04:50:27FromDiscord_<aolko> oh, it can't
04:50:35FromDiscord_<aolko> ouch
04:50:53*vlad1777d quit (Ping timeout: 245 seconds)
04:51:39rayman22201weird thing to say. You can probably make bindings for it.
04:51:50rayman22201It's a small community, so it probably hasn't been done yet.
04:52:09FromDiscord_<aolko> not as small as rust's
04:52:17*Tyresc quit (Quit: WeeChat 2.4-dev)
04:52:26rayman22201what? It's much smaller than Rust
04:52:34FromDiscord_<aolko> ouch
04:53:09rayman22201ouch?
04:53:23FromDiscord_<aolko> oof with taste, if you will
04:54:05rayman22201popularity is not the judge of good programming languages.
04:54:32FromDiscord_<aolko> i know, yet the bigger community is, the more tutorials and outreach people get
04:54:48FromDiscord_<aolko> the more researched the language becomes
04:54:56rayman22201That just takes time, and we have no big corporate backing
04:55:02rayman22201I point you here: https://forum.nim-lang.org/t/4510
04:55:03FromDiscord_<aolko> the more ordinary tasks it get used for
04:55:52FromDiscord_<aolko> https://i.imgur.com/b33TR86.png
04:55:55rayman22201If you are looking to troll, then we don't have anything to discuss. If you are looking for a legitimate discussion of the pros and cons of our langage, I will do my best to help.
04:56:19FromDiscord_<aolko> not trolling, 100% genuine, you detector is broken, abort abort
04:56:24FromDiscord_<aolko> your*
04:57:14FromDiscord_<aolko> i'm just poking around, wondering if i will regret learning nim from someone from here
04:57:21*dddddd quit (Remote host closed the connection)
04:58:02FromDiscord_<aolko> nim by example, as they say, starting with writing sugar macros
05:01:47rayman22201The best way to do the hash table / "array" thing you are describing would be something with variant types. Or something like the jsonobject from jsffi: https://nim-lang.org/docs/jsffi.html
05:02:16shashlickCheck out tables, is a good start
05:02:31shashlickIf you need more sugar, we can certainly look into it
05:03:22shashlickMost of the time, you will get questions about your design if you ask for things common in interpreted languages though
05:03:47shashlickPoint is interesting though, a data structure that is more flexible
05:06:01FromDiscord_<aolko> tl;dr i need arrays from php in nim, or at least from ruby, that's split them in two - arrays and hashes
05:06:13rayman22201Variant types aren't flexible enough? Whenever I have ever used "tagged hash maps" or dynamically typed languages like that (and I have used everything from lisp to php to python), that structure ends up being a stringly typed mess that causes more bugs than it solves. I want a strictly typed language for a reason... so I can describe my probrem domain with the types.
05:08:14shashlickYou could have an array of pointers
05:08:18rayman22201In those languages, the implementation is always a hash table behind the scenes. It's never actually an array. When you are doing array[0] in php it's really using "0" as the key to a hash map.
05:08:25shashlickAnd cast back and forth
05:08:38FromDiscord_<aolko> ¯\_(ツ)_/¯
05:10:09rayman22201table is definitely what you want then: https://nim-lang.org/docs/tables.html
05:13:05rayman22201Along with this: https://nim-lang.org/docs/manual.html#types-object-variants
05:13:49FromDiscord_<aolko> iirc tables are from lua
05:14:21FromGitter<sotrhRaven> First troll of the year.
05:14:30rayman22201table is what Nim calls hash map. I think lua does the same
05:15:08FromDiscord_<aolko> people and their faulty troll detectors, *sigh*
05:15:47FromDiscord_<aolko> rayman22201, okay, that's one bit of an array, now it needs to be merged
05:17:27rayman22201You do come off a bit harsh. Just saying... I don't come into your home and tell you that your oven is awful and that you should use an over that works like my oven.
05:18:01rayman22201You would "merge" it, but using a variant type as the value of your hashmap
05:18:52rayman22201Nim is about Algebraic Data Types. Again, closer to Rust, Ada, or Haskell
05:20:46FromDiscord_<aolko> yes, pardon me, i'm just still hot from rust arguing
05:22:22rayman22201I can appreciate that. Nobody likes the Rust evangilist strike force... I've met religious fanatics that are less stubborn...
05:22:30*Senketsu quit (Quit: Leaving)
05:23:10shashlickSo static languages are more strict with types
05:23:40FromDiscord_<Generic> the things is, in dynamic programming languages you usually already have some kind of typing implicitely in your mind
05:23:42FromDiscord_<aolko> i just want that sugar macros with all the familiar things i know and like
05:24:18FromDiscord_<Generic> you know, this function takes numbers, this function takes a table with that key of that type
05:25:20FromDiscord_<Generic> in a programming language with static types you have to formulate these types for the compiler
05:25:41FromDiscord_<Generic> and you're going to be rewarded by far better performance
05:26:21FromDiscord_<aolko> sure but what if i want the types to be figured out by the compiler, besides in nim those are optional
05:26:35FromDiscord_<Generic> they aren't
05:26:35shashlickTypes are mandatory in Nim
05:26:53FromDiscord_<Generic> if you provide a value, the compiler figures out which type it has
05:27:32FromDiscord_<Generic> it just saves you typing and it looks cleaner
05:27:38rayman22201Types are required in Nim. Nim has a smart compiler that can figure out the types automically sometimes.
05:27:57FromDiscord_<aolko> https://learnxinyminutes.com/docs/nim/ cheatsheet says you can declare value w/o types
05:28:37FromDiscord_<Generic> that doesn't mean it has no type, it just advises the compiler to infer it
05:28:54rayman22201But you have to give types in other places such as function arguments
05:29:09FromDiscord_<aolko> well, that's probably going to be a second macro, if it can be done w/it
05:29:38FromDiscord_<Generic> as I said above, you usually know what arguments you expect to be given to a function
05:29:52rayman22201Here is project that does runtime variant types (what you are asking for) https://github.com/yglukhov/variant
05:29:58FromDiscord_<Generic> screw usually, always!
05:30:12*Ven`` joined #nim
05:30:20FromDiscord_<Generic> if you don't know what kind of data you're function is working on, you're unable to write it at all
05:31:28FromDiscord_<aolko> okay, can that variant be "embedded" more tightly?
05:31:42FromDiscord_<aolko> w/o having to call it every time
05:31:43rayman22201I can't vouch for the quality of that project. It hasn't been updated super recently. Yglukhov produces good code generally though.
05:31:59FromDiscord_<Generic> no, and that's good that way
05:32:48rayman22201You could write a macro to do a lot of the boilerplate for you. Tables + runtime variants + macros, would give you what you want aolko
05:32:57FromDiscord_<aolko> `var v = newVariant(5) (int)` < `var v = 5 (int)`
05:33:04FromDiscord_<Generic> maybe give us a small daily live example and we show you how to think with explicit types?
05:33:16FromDiscord_<aolko> or just `v = 5` (python/ruby style)
05:33:56FromDiscord_<aolko> which live example?
05:34:42FromDiscord_<Generic> I don't try to think of something that's not possible with explicit types, instead just think of some programming problem
05:34:45FromDiscord_<aolko> as of sugar you can look at moonscript and coffeescript, besides ruby
05:34:55FromDiscord_<aolko> as of sugar you can look at moonscript and coffeescript, besides ruby and crystal
05:35:34FromDiscord_<Generic> Nim isn't a scripting language, it only has a pythonesque syntax, but otherwise it's closest relatives are Pascal, C++ and maybe Lisp
05:35:38FromDiscord_<aolko> okay, how about this one https://github.com/aLingua/pphp/blob/master/spec/syntax.md
05:36:44*Ven`` quit (Ping timeout: 246 seconds)
05:37:02FromDiscord_<aolko> i need a toolbox for that tho, which is custom grammar notation (PEG-like), parsergen, compiler and all that in nim
05:37:42FromDiscord_<aolko> sure it's now a webbrowser or a game, but that's one hot project
05:37:53FromDiscord_<aolko> not*
05:38:20rayman22201Nim has a Peg parser, and all that
05:38:44rayman22201https://nim-lang.org/docs/pegs.html
05:38:51FromDiscord_<aolko> yes, yet the toolbox should be useable by everyone
05:39:07FromDiscord_<aolko> even noobs
05:39:22FromDiscord_<aolko> to improve and hack in the language
05:40:15FromDiscord_<aolko> that means, very sugary, very simple syntax in each and every part of the toolbox and lots of comments
05:40:51FromDiscord_<aolko> as of notation, why peg like, i think peg can be further simplified
05:41:05rayman22201I find Nim's syntax very simple, but I also come from years of C++ lol
05:41:15FromDiscord_<Generic> me too
05:41:21FromDiscord_<aolko> see:http://cjheath.github.io/treetop/
05:41:38FromDiscord_<aolko> a bit simplier than peg notation
05:43:00FromDiscord_<Generic> afaik the parser of Nim is hadnwritten
05:43:26rayman22201https://nim-lang.org/docs/pegs.html#eventParser.t%2Cuntyped%2Cuntyped
05:43:41rayman22201That seems pretty clear imo
05:44:15rayman22201But also, yes, the Nim parser itself is handwritten and has lots of performance optimizations
05:44:32FromDiscord_<aolko> granted, not as simple as https://hastebin.com/ejuwolekez.nginx, let's admit
05:46:25rayman22201True. This is going to sound like I'm trolling, but you could probably make a macro to make Nim's Peg module look like that though. Nim's macros are on par with Lisp macros in power.
05:47:09rayman22201The thing is, we generally don't go that far, because we want to see the inner workings to have more control over performance etc...
05:47:30FromDiscord_<aolko> wait, this one is not for nim macro, this is a separate notation, need a parsergen and a compiler tho
05:48:03rayman22201I'm saying you could use a nim macro to simplify the syntax of the Nim parser gen
05:48:38rayman22201If you want to see a full compiler though, you can see the Nim compiler itself. Nim is written in Nim.
05:49:13FromDiscord_<aolko> ...without going for nim internals?
05:49:33FromDiscord_<aolko> just starting from scratch, except the parser, due to parsergen
05:50:31rayman22201Nim compiler maybe a bad example. too complex. Karax is a good example of Nim macros being used to make simple and powerful DSLs: https://github.com/pragmagic/karax
05:50:59FromDiscord_<aolko> not a dsl either, sorry
05:51:23FromDiscord_<Generic> the Nim compiler actually doesn't use many macros
05:51:58FromDiscord_<Generic> nim macros still have to follow nim syntax rules
05:52:04FromDiscord_<aolko> nim is just a platform for said language, it does not extend nim in any way
05:52:06rayman22201I think maybe I don't understand your question aolko.
05:52:23FromDiscord_<aolko> it's like crystal, it was written on ruby
05:52:32FromDiscord_<aolko> then it got rewritten in crystal
05:52:49FromDiscord_<aolko> same here, this pphp is getting written in nim
05:53:00FromDiscord_<aolko> possibly to be rewritten in itself
05:53:09FromDiscord_<aolko> bootstrapping, y know
05:54:36FromDiscord_<Generic> sorry, I think, I also don't quite understand what want to know
05:54:51rayman22201Something like this? https://min-lang.org/
05:55:26rayman22201Min is a dynamic scripting language that was written in Nim as the underlying runtime.
05:55:48FromDiscord_<aolko> eh kinda
05:56:03FromDiscord_<aolko> @Generic okay, step-by step
05:56:10FromDiscord_<aolko> we grab nim, right
05:56:44FromDiscord_<aolko> we use a parsergen to make a parser for our custom peg-like notation
05:56:56FromDiscord_<aolko> then we write the interpreter in nim
05:57:09FromDiscord_<aolko> then maybe a compiler...still in nim
05:57:18FromDiscord_<aolko> then we have pphp
05:57:32FromDiscord_<aolko> then we probably rewrite pphp in pphp
05:57:41FromDiscord_<aolko> kinda like that
05:58:20FromDiscord_<aolko> but given the fact pphp, is a web language we might not need a compiler
05:58:35FromDiscord_<aolko> instead we'll need some sort of a server, like apache
05:58:38FromDiscord_<Generic> so you're goal is to bootstrap pphp's developement by writing an interpreter for it in Nim?
05:58:41FromDiscord_<aolko> so there's that
05:58:57rayman22201I think I follow now. Honest question, if you want to bootstrap eventually, why not use a language you are more familiar with as a starting point?
05:59:04FromDiscord_<aolko> yes, if pphp existed beforehand
05:59:11FromDiscord_<aolko> it does not exist yet
05:59:16FromDiscord_<Generic> pphp is going to be an interpreted programming language?
05:59:28FromDiscord_<aolko> web one, better php
05:59:34FromDiscord_<aolko> web one, better/simpler php
06:00:09FromDiscord_<aolko> because:
06:00:09FromDiscord_<aolko> ruby and python are p bad at being a language platforms
06:00:17FromDiscord_<aolko> and i don't know haskell
06:00:26FromDiscord_<Generic> if that's the case it's not really possible somewhen write the interpreter in that language
06:00:45rayman22201lol. Haskell is good at what you are trying to do, this is true. But Haskell is also.... Haskell....
06:00:51FromDiscord_<aolko> then it will remain in nim, w/o further bootstrapping
06:01:16FromDiscord_<aolko> haskell is weird, never got it
06:01:29FromDiscord_<aolko> so the next best thing is nim
06:01:47rayman22201Haskell is weird. I agree. I even kind of like Haskell, but it is definitely weird.
06:01:48FromDiscord_<Generic> you can't write an interpreter in the same interpreter language, without keeping the old interpreter
06:01:48FromDiscord_<Generic> that interprets the new interpreter
06:01:49FromDiscord_<Generic> the next best thing would be Ocaml
06:02:07FromDiscord_<aolko> excuse me, i have a violent allergies to brackets
06:02:45rayman22201Nim does use brackets for generics. That may bother you also :-P
06:02:57FromDiscord_<aolko> regular ones, not curly ones
06:03:14rayman22201true that. Nim is also alergic to curly braces
06:03:32FromDiscord_<Generic> they are used in the language
06:04:03FromDiscord_<Generic> for term rewriting macros
06:04:05*kapil____ quit (Quit: Connection closed for inactivity)
06:04:05rayman22201True.
06:04:07FromDiscord_<aolko> anyway, how do i start with the peg-like notation?
06:04:30FromDiscord_<Generic> do you experience with any strongly typed programming language?
06:04:41FromDiscord_<Generic> C++, Java, Typesrcipt?
06:04:52FromDiscord_<aolko> no, golo>java, no
06:05:43rayman22201but back to your question. Min-lang is actually pretty similar to what you are trying to do. It is definitely possible to do what you are trying to do in Nim.
06:06:01FromDiscord_<aolko> and i know that
06:06:07rayman22201The pegs link I sent you is a good starting point
06:06:30FromDiscord_<aolko> but the end result is the hastebin link i sent
06:06:52FromDiscord_<Generic> although a hand written parser would be much more performant 😃
06:08:10FromDiscord_<Generic> ok, that's coming from the person who wrote nothing more than a S-exp parser in his life
06:08:35rayman22201You aren't going to get something as clean as that paste-bin peg syntax without doing quite a bit of work. It's not built in.
06:09:01FromDiscord_<aolko> not denying that
06:09:39FromDiscord_<Generic> the thing is, probably nobody has used he peg module yet to something at this scale
06:10:06FromDiscord_<aolko> it's peg-***like***, not peg
06:10:24FromDiscord_<Generic> if you write your own macro to generate the parser it probably takes the same time as writing a handwritten parser
06:10:53FromDiscord_<Generic> although it's going be less easy to maintain
06:11:41rayman22201Wait, so you have a language, pphp, but you also have your own "peg like" grammar to describe the language?
06:12:30FromDiscord_<aolko> no, first goes the egg, then the chicken
06:12:46FromDiscord_<Generic> if the language is php like, it should probably have very fast startup times?
06:12:56FromDiscord_<aolko> we have a custom notation ***of pphp*** that describes it's grammar
06:13:08FromDiscord_<aolko> parsergen shall start from here
06:13:17FromDiscord_<aolko> i.e. grab that notation
06:13:51FromDiscord_<aolko> if you've seen the spec it's a cross of ruby and php
06:14:54FromDiscord_<Generic> in theory it's possible to write a parser that takes such a grammar file and generates an appropriate parser
06:15:16FromDiscord_<aolko> yup
06:15:18FromDiscord_<Generic> but again, esp. if you want very fast startup times, you probably even want to generate an AST while parsing
06:15:45FromDiscord_<aolko> that includes the ast and whatever is required
06:16:16FromDiscord_<Generic> I'm repeating myself, but a handwritten parser is the way to go
06:16:28FromDiscord_<Generic> handwritten doesn't mean it has to be ugly
06:16:54FromDiscord_<Generic> this isn't C nor C++
06:16:58FromDiscord_<aolko> and it also means i will not be able to doit on my own
06:17:28FromDiscord_<aolko> entirely at least
06:17:37rayman22201It's a pretty ambitious project to use as your first project in a language you have never used before.
06:17:58FromDiscord_<aolko> duh, i said before it's ***hot***
06:18:06rayman22201lol
06:18:22FromDiscord_<aolko> and i kinda want to replace php
06:18:33FromDiscord_<aolko> getting sick and tired of that oop
06:18:35FromDiscord_<Generic> https://en.wikipedia.org/wiki/Recursive_descent_parser#C_implementation
06:19:36rayman22201I get that. Good goals. But I would do a few smaller things in the language first to get a feel for the language first.
06:19:47FromDiscord_<aolko> no time sadly
06:20:01FromDiscord_<aolko> no time to waste on little things
06:20:27FromDiscord_<Generic> if you're thinking that way, you're probably going to get anywhere
06:20:36FromDiscord_<Generic> *nowhere
06:21:42FromDiscord_<aolko> pardon but please spare me that, just want to carry on with the goal
06:21:50rayman22201To give an analogy: You are asking to write a AAA video game like Crysis when you don't even really know how to make openGL calls yet... It's kind of tough
06:22:35FromDiscord_<aolko> *you actually have a perfect cryengine for that*
06:22:42rayman22201They write books on how to write compilers and interpeters. It's not something you just jump into
06:23:10FromDiscord_<aolko> that's why i said i won't be able to do that entirely on my own
06:23:21rayman22201exactly. You are basically asking how to write the equivalent of cryengine.
06:23:54*kapil____ joined #nim
06:25:23FromDiscord_<aolko> ehhhh, wouldn't really say that, if you have such tools available - sure, let's get those running, if you don't let's make a barebones required versions of those
06:25:42FromDiscord_<aolko> if nim'd had teetop analogue/port, i'd use that
06:25:48FromDiscord_<aolko> maybe
06:26:40rayman22201The closest thing to treetop is the Nim Pegs module.
06:27:07FromDiscord_<aolko> yet grammar it offers is just raw peg
06:27:12FromDiscord_<aolko> which sucks a lot
06:27:23FromDiscord_<Generic> yeah, but that's the best there is
06:27:29rayman22201That's the state of Nim atm
06:27:36FromDiscord_<Generic> than don't use Nim
06:27:37rayman22201again, contributions welcome.
06:28:04FromDiscord_<aolko> sure let me get my delphi running
06:28:54rayman22201lol. cool. go for it.
06:29:20FromDiscord_<aolko> i hope nim's sources are heavily commented
06:29:35FromDiscord_<Generic> you mean the compiler sources?
06:29:47FromDiscord_<aolko> all of them, even grammar
06:30:16rayman22201Nim is pretty comprehensible imo but it's big.
06:30:20FromDiscord_<Generic> they probably shoudnl't be that interesting too you, since you're writing an interpreted language
06:30:26rayman22201check out Min. seriously
06:30:30rayman22201it's a good example for you
06:30:55FromDiscord_<aolko> min is meh, i'd say it's far from where i'm heading
06:31:20rayman22201Ignore all the functional programming crap. Just look at the parser. It's a good example of a handwritten parser in nim
06:31:40rayman22201You have to separate the useful parts from the not useful parts.
06:31:56FromDiscord_<Generic> if you want something more basic, look at the standard library json parser
06:33:28FromDiscord_<aolko> even in the parser there's lot's of crap, references on top of references, geez
06:33:57FromDiscord_<Generic> complexity goes in many directions
06:34:01FromDiscord_<aolko> and just a couple of comments like `##this is a variable, lol`
06:34:14FromDiscord_<Generic> than look at the json parser or the xml parser
06:35:41Araqactually the parser has #| comments that give the parsed syntax in ABNF and there is no need for comments because it's not for teaching you the basics of parsers, there are books for that.
06:35:52FromDiscord_<aolko> better but still not yet as obvious
06:38:26Araqseriously, go read a book.
06:38:44rayman22201You may appreciate something like ANTLR aolko
06:39:02rayman22201https://www.antlr.org/
06:39:04FromDiscord_<aolko> no i didn't and i tried to
06:41:06rayman22201look, you are asking for something very hard, and you don't seem to want to put the work into properly doing it. There is no language that I know of that is going to be both high performance / basis for a platform, and "fit your exact definition of easy" to make parsers.
06:41:38rayman22201We are giving you some starting points, but we aren't going to teach you how to write a compiler.
06:43:02FromDiscord_<aolko> may i ask who's willing to? I need cut-to-the-chase no-bs tl-dr explanation on ***howto*** so it can be implemented in practice afterwards
06:43:13AraqHint: Parsing is the easy part in a compiler. ;-)
06:43:42FromDiscord_<aolko> don't need one, it's interp language
06:44:14FromDiscord_<Generic> that only makes it a bit easier
06:44:24rayman22201Are you asking for volunteers, or are you asking for someone to hire? We are already volunteering our time to a language... Nim... and our time is valuable. If you want to hire somebody, make a bounty on bounty source or post an ad on the forum.
06:44:36FromDiscord_<aolko> it's an open source project
06:44:48rayman22201so is Nim
06:44:49FromDiscord_<aolko> no payments for hiring
06:46:31FromDiscord_<aolko> if i'd wanted to commission myself a language, i'd done that long ago
06:47:45FromDiscord_<Generic> I actually also wanted to write my own programming languages in my most frustrated days of C++
06:47:56FromDiscord_<Generic> but then I learned about Nim 😃
06:48:01rayman22201Why would you think this is a good place to recruit people? Again. Everyone here already has found a language we like... If we didn't we wouldn't be here.
06:48:31FromDiscord_<aolko> because it will be done ***in nim***
06:48:39rayman22201We already think Nim is a better Php, and a better Python, and a better Javascript, and a better C++
06:49:19rayman22201I'd rather just contribute to Nim and make Nim better.
06:51:25rayman22201It's a small community. Make a forum post. Maybe somebody will be interested. I wouldn't hold my breath though. You should be more upfront with your intentions though. If you want to recruit somebody, say so, don't come in pretending that you are trying to do a project yourself.
06:51:57FromDiscord_<Generic> without an understanding of idiomatic Nim and the theory behind interpreters, your pphp interpreter in Nim will end up similar like my failed programming languages
06:52:15FromDiscord_<aolko> i'm not pretending, i've outright said that i won't be able todo it all by myself
06:53:01FromDiscord_<Generic> then you're not ready to take this task
06:53:25FromDiscord_<Generic> even if other people do some work, you still have to grasp how and what they are doing
06:53:37FromGitter<alehander42> aolko, why don't you sit down and study a bit indeed, I am sick of this
06:54:03FromDiscord_<aolko> I need cut-to-the-chase no-bs tl-dr explanation on howto so it can be implemented in practice afterwards
06:54:23FromDiscord_<Generic> there is nothing like that
06:54:39FromGitter<alehander42> if you can't learn how to write a simple language with today's internet resources, sorry, but you are not really good enough to create one
06:55:13FromGitter<alehander42> go, read e.g. http://craftinginterpreters.com/
06:55:17FromGitter<alehander42> and then come back
06:55:20FromDiscord_<aolko> in turn you are not good enough to tell people that they are not good enough for anything
06:55:37FromGitter<alehander42> oh, I know your case, please don't get me started
06:56:00FromGitter<alehander42> how many years does it take to stop being lazy and sit down and read at least a simple compiler text?
06:56:29Araqhttps://techxplore.com/news/2018-10-artificial-intelligenceparking-car-neurons.html offtopic but I'm sure #nim appreciates it
06:56:43rayman22201alright. No need for personal attacks. Aolko, I don't think we can help you in the way that you are looking for. Good luck to you.
06:56:48FromDiscord_<aolko> how many years it will take to dunk the head in a bucket of water and actually get anything?
06:57:27FromDiscord_<aolko> because that's what those huge manuals/books are, a bucket of water
06:57:35rayman22201@namiran will love that for sure Araq.
06:57:53FromDiscord_<Generic> I was in a similar situation like that not to long ago
06:58:05FromGitter<alehander42> @aolko there are hundreds of simpler manuals/texts/blogs/etc. langdev is probably one of the *most* researched/blogged/teached areas ever
06:58:15FromDiscord_<Generic> I wanted to write a JIT compiler backend for a console emulator
06:58:33FromDiscord_<Generic> it took me four attempts to get it right
06:59:06FromDiscord_<Generic> it's embarassing how much time you can spend in reading processor manuals
06:59:18FromDiscord_<aolko> those tiny manuals are fixed on concrete examples. "Make a lisp". Nope. "Make a compiler in js". Nope. "Make a basic interpreter". Nope again.
06:59:59FromDiscord_<Generic> than why don't look the implementation of other interpreted languages?
07:00:07FromDiscord_<aolko> i need a manual that is chained to that very case
07:00:09FromGitter<alehander42> but this is how you learn, aolko. and
07:00:24FromGitter<alehander42> there is not a simpler way
07:00:40FromDiscord_<Generic> there is no manual, if there were one, someone would have already written exactly that software you want to write!
07:00:43FromGitter<alehander42> there is no magical tutorial that solves your own design case
07:00:54FromDiscord_<aolko> `than why don't look the implementation of other interpreted languages?` because those are huge uncommented steaming pile of hot garbage
07:01:26FromDiscord_<aolko> look at ruby's parser for example https://github.com/ruby/ruby/blob/trunk/parse.y
07:01:42FromDiscord_<aolko> the only thing i learned from there's it's damn huge
07:01:56FromGitter<alehander42> and please, stop being arrogant, you don't really know much about the field, calling other language pile of hot garbage just of pure ignorance is not nice.
07:02:11Araqread the book about "compiler construction in Oberon"
07:02:26FromGitter<alehander42> it's trivial to tweak a small language from another tutorial, once you know how it works, to do what you want to do
07:02:26*FromGitter quit (Read error: Connection reset by peer)
07:02:40rayman22201Nobody has a manual for "the exact thing you want". Extrapolating principles and patterns from a concrete example is one of the core tenants of learning and engineering.
07:02:45*FromGitter joined #nim
07:03:11FromDiscord_<aolko> alright, plan b, explain the material like you would to a drunk 5 y/o
07:03:51rayman22201http://craftinginterpreters.com/ is exactly that
07:04:24FromDiscord_<aolko> it's unfinished
07:04:30FromGitter<alehander42> it doesn't matter
07:04:37FromGitter<alehander42> it has all you need for your prototype
07:04:50FromDiscord_<Generic> that's sometimes simply not possible
07:04:50FromDiscord_<Generic> or it is, the first layer which allows you to go deeper and deepr
07:05:24FromDiscord_<aolko> besides it uses things that are thrown away in peg grammars
07:05:33FromDiscord_<aolko> the need for scanning etc
07:06:16FromGitter<alehander42> aolko, most real languages use handwritten parsers
07:06:32FromGitter<alehander42> but again, this doesn't matter: you can skip the parser part and just use a parser generator
07:06:47*thumbs_scum joined #nim
07:06:48Araqhttps://github.com/lua/lua/blob/master/lparser.c#L945
07:08:43*narimiran joined #nim
07:09:38*narimiran quit (Remote host closed the connection)
07:11:10*narimiran joined #nim
07:19:03*sz0 joined #nim
07:25:03FromGitter<alehander42> aolko: another relatively eli5 one: https://ruslanspivak.com/lsbasi-part1/
07:52:07*xace quit (Remote host closed the connection)
08:34:04*kapil____ quit (Quit: Connection closed for inactivity)
08:34:17*xet7 joined #nim
08:41:03*ChristianWitts joined #nim
08:43:11narimirananybody wants to take a shot at benchmarking nim vs cpp, rust and dlang? https://atilanevesoncode.wordpress.com/2018/12/31/comparing-pythagorean-triples-in-c-d-and-rust/
08:46:42Araqbe my guest but anything you measure is implementation artifacts, on paper all of them are on par with value semantics, LLVM based backends etc yada yada
09:27:21*sz0 quit (Quit: Connection closed for inactivity)
09:35:42*gangstacat joined #nim
09:40:45*Vladar joined #nim
09:41:19*Pisuke joined #nim
09:41:26*MyMind quit (Ping timeout: 268 seconds)
09:43:50*stefanos82 joined #nim
10:09:49Araqyay my new year's hack, works
10:10:29Araqblog post worthy stuff, an alternative to exceptions in a 20 line compiler diff
10:10:47narimiranwow, nice
10:12:23Araqit's ridiculously simple, probably why this solution has been overlooked for so long
10:13:53Araqit's also wrong of course, but my bet is we'll be surprised how far this idea can be taken
10:25:44FromGitter<alehander42> what does it look like?
10:28:49*nsf joined #nim
10:41:39*vlad1777d joined #nim
10:48:27*ChristianWitts quit (Ping timeout: 240 seconds)
11:09:59AraqI won't spoil the content of my blog post
11:19:58FromDiscord_<exelotl> Nim 0.20 will have a battle royale mode
11:48:41Araqchoosenim supplies 0.19.0 ?
11:48:45Araqcan we fix that?
11:49:51*kapil____ joined #nim
11:54:04*dddddd joined #nim
11:55:32narimiranAraq: even if we discard that, choosenim really needs a version bump
11:57:12narimiranping dom96 ^
12:01:05*nc-x joined #nim
12:01:42nc-xWhats with tests/stdlib/tmath.nim? It keeps randomly failing on CI.
12:02:00nc-xBut passes locally.
12:03:52nc-xFor eg https://ci.appveyor.com/project/dom96/nim/builds/21316407/tests
12:04:02*nc-x quit (Client Quit)
12:08:54narimirannc-x: i just skimmed over your #10151, so i'm wondering why we have twice 'is deprecated' here: "the '.this' pragma is deprecated is deprecated" ?
12:09:32narimiran(line 1104 of compiler/pragmas.nim)
12:11:54dom96<Araq> choosenim supplies 0.19.0 ?
12:11:56dom96It doesn't?
12:12:18narimirandom96: see the comment here: https://forum.nim-lang.org/t/4512#28211
12:12:30narimiran(and please delete mine after it, i cannot delete it!?)
12:12:49*nc-x joined #nim
12:12:56nc-xnarimiran: oops
12:14:57Araqdom96: ok, my bad, sorry
12:15:31nc-xfixed now
12:15:49dom96I don't see why that user had 0.19.0 install
12:16:12dom96oh, probably because they already have choosenim...
12:17:02narimirannc-x: great. since Araq has already given you his comment, i think this is safe to merge (any new thoughts Araq? https://github.com/nim-lang/Nim/pull/10151)
12:17:35Araqnc-x: I dunno, tfloatmod also fails in C++ mode even though targets lists "C++" explicitly
12:19:00dom96Araq: This is why we include instructions on how to update in the release notes. It would be helpful to include that in your forum post too
12:19:34dom96(I already edited it to include a link)
12:20:38Araqok ty
12:22:13dom96Time to compose a tweet about the release
12:22:48narimirandom96: i can do that, so you can have more time for releasing new choosenim ;)
12:23:28narimiranbtw, i'm waiting with reddit/HN post about the release — lots of people will be hangover today
12:23:34dom96That would take far longer than composing a tweet :P
12:23:51narimirandom96: i can compose it very very slowly :P
12:24:07dom96Loving this forum bug where it shows all threads from last year as "Dec 2018" :D
12:24:48dom96But okay, go for it, here is a template for you: https://twitter.com/nim_lang/status/1045305715361865728 :)
12:25:24dom96There may or may not be a correlation between the amount of emojis you use and the amount of likes the tweet gets :D
12:25:41dom96So why do we need a new choosenim release?
12:28:27narimiranthis is "nimpretty release", so it would be nice to have "Add nimpretty to the list of installed tools.", among others
12:29:23dom96It seems I'm still waiting on this https://github.com/dom96/choosenim/pull/63 :/
12:29:53narimiranyeah, that would indeed be nice
12:30:19dom96I'll see what I can do, but no promises. I've got other stuff I want to work on (last day of holidays D:)
12:30:28Araqdom96: merge and touch it, it's what I do too
12:30:55Araqyou know the saying, "If you want something done, do it yourself"
12:36:46*natrys joined #nim
12:39:09*Perkol joined #nim
12:39:45PerkolHow do I send request in httpclient without cookies
12:39:47Perkol?
12:44:39FromGitter<dom96> Without?
12:44:55FromGitter<dom96> It doesn't send cookies unless you specify them
12:45:13FromGitter<dom96> Araq: heh yeah. I'd need to break out my Windows PC for that though
12:47:22narimirannc-x: iirc #10151 was passing CI before your force-push. right?
12:50:35nc-xnarimiran: yes
12:50:58narimiranmerged
12:53:09*narimiran quit (Quit: Leaving)
12:59:04*stefantalpalaru joined #nim
13:00:00*Perkol quit (Remote host closed the connection)
13:00:12stefantalpalaruSomeone please update the csources to 0.19.2 - https://github.com/nim-lang/csources. Bonus points for automating it as part of the release process.
13:00:36Araqstefantalpalaru: why is it required?
13:01:04AraqI deliberately left it out, old C sources can build 0.19.2 just fine
13:06:23Araqin fact we should ensure the C sources are never touched anymore, then the code can be audited for that unrealistic "Thompson hack"
13:10:25stefantalpalaruMy Gentoo ebuild bootstraps the compiler from the same csources version. Why sabotage this kind of automation?
13:11:28*nc-x quit (Quit: Page closed)
13:16:10*krux02 joined #nim
13:40:08*kidandcat[m] joined #nim
13:55:41*natrys quit (Quit: natrys)
14:01:00*Ven`` joined #nim
14:19:00*Tyresc joined #nim
14:24:28thumbs_scumwhy not automate sabotage?
14:27:34*Ven`` quit (Ping timeout: 268 seconds)
14:31:43Zevvactually *implementing* a thompson hack would be pretty slick
14:32:08Zevva thompson easter egg, just because you can
14:32:08*Ven`` joined #nim
14:43:55FromGitter<sheerluck> @Araq @stefantalpalaru it would be nice to have csources-0.19.2 bc this: https://dumpz.org/cxRdBPqT8NfP
14:46:41stefantalpalaruI'm doing it with a CSOURCES_VERSION variable, but it's an avoidable complication keeping track of two different versions for one package. Avoidable complexity is the root of all evil.
14:51:29dom96huh, we always host C sources on the website... https://nim-lang.org/download/nim-0.19.2.tar.xz
14:51:35dom96Why not use that?
14:51:47Araqkeep the CSOURCES_VERSION on 0.19.0 for good?
14:52:03Araqwhat never changes doesn't have to be "automated"
14:54:14stefantalpalaruHow long until you are forced to update the csources because the compiler uses some new language feature?
15:08:47stefantalpalaru@dom96 - thanks, I'll switch to that archive with bundled csources.
15:08:48*kapil____ quit (Quit: Connection closed for inactivity)
15:12:05*krux02 quit (Ping timeout: 250 seconds)
15:13:07stefantalpalaru`koch test` is broken because it looks for tester.nim in testament/ instead of tests/
15:20:54leorizestefantalpalaru: it's not broken
15:21:10leorizethe archive just doesn't have testament included
15:24:59stefantalpalaruThat's the spirit! Now let's get someone to fix this "feature" in a 0.19.3 release.
15:26:24*nc-x joined #nim
15:26:27leorizeI believe the devs consider testament a development-only thing
15:26:42leorizeso it makes sense that it's not in the "release" archive
15:26:51nc-xstefantalpalaru: why do you think this should be fixed? testament is for developers of the language not users of the language.
15:27:28stefantalpalaruAre you seriously asking why I want to be able to run the test suite in an official release of the compiler?
15:27:34nc-xYes
15:27:49nc-xThe release is tested on the CIs before being released
15:27:58nc-xwhy should the tester be included then?
15:28:00stefantalpalaruObviously not.
15:29:19*xace joined #nim
15:30:15leorizestefantalpalaru: can you move the nim config folder in your ebuild to /etc?
15:30:19stefantalpalaruWhoever cherry-picked commits missed the tester.nim move, probably by not running the test suite from the resulting package. This means that you've published a partially broken release. This is not something to hand-wave if you care about software quality.
15:30:47leorizehttps://github.com/nim-lang/Nim/tree/v0.19.2
15:30:57stefantalpalaru@leorize, I can, if it's still picked up in there.
15:30:59leorize^ what if I told you, it was moved long ago?
15:31:46leorizestefantalpalaru: it should be put in /etc/nim
15:32:29leorizealso, can you left the $lib/compiler portion out of the config? it should be used as `import compiler/<module>`, not `import <compiler module>`
15:32:37leorizeit conflicts with the options module
15:33:33stefantalpalaruRunning the testsuite with koch is part of the official documentation, not some developer-only detail: https://github.com/nim-lang/Nim/#koch
15:35:09nc-xstefantalpalaru: As can be seen in the link by leorize, the CIs run and are green, so it is very well tested.
15:35:25nc-x`koch is the build tool used to build various parts of Nim and to generate documentation and the website, among other things. `
15:35:38nc-xWhere does it mention that koch should be used by `users` of the language?
15:35:40dom96The CI doesn't execute `koch tests`
15:35:41dom96https://github.com/nim-lang/Nim/blob/v0.19.2/.travis.yml#L46
15:36:15thumbs_scumkoch pronounced like crotch?
15:36:35nc-xdom96: Ah, okay...
15:37:03dom96But yeah, the test suite isn't intended to be run by users or packagers
15:37:41*nc-x quit (Quit: Page closed)
15:38:14stefantalpalaruAbsurd.
15:43:46leorizestefantalpalaru: you are right, someone didn't backport the commit that ship testament in release tarball https://github.com/nim-lang/Nim/commit/00f84d3d22cdeed7b4a2060e6307e3a742729055
15:44:02leorizewell, mainly because Araq didn't tag it
15:44:30thumbs_scumso its a new testament?
15:48:29stefantalpalaruThe bigger problem is that those green CI tests are not testing the resulting archive. What happens in compiler/installer.ini is completely untested.
15:55:50stefantalpalaruI'd insert archive creation and archive unpacking in .travis.yml followed by a "cd" to run all the tests in what the end user receives.
15:59:52FromGitter<dom96> AFAIK that's the responsibility of the new nightlies repo. Yes, it's still quite immature. Please make issues on it.
16:06:41*nsf quit (Quit: WeeChat 2.3)
16:08:30Araqno idea why the move to testament/ got backported incompletely.
16:13:14Araq"I'd insert archive creation and archive unpacking in .travis.yml followed by a "cd" to run all the tests in what the end user receives."
16:13:28Araqwe used to do that but travis timed out with that ;-)
16:14:20ZevvI'm trying to figure out why the `==` proc is listed 27 times in the exports section of the nativesockets docs. I was able to go deep enough to see this seems to happens in the sem pass, but I'm getting lost here.
16:15:01Araqhe he
16:16:04ZevvThe PNode going into the sem pass has == only once, but it comes out 27 times. I'll file an issue and go fix something easier :)
16:16:35Araqbut the question remains, what does the Linux packager do with a package that has red tests. Not update it?
16:17:06AraqI bet every single release of GCC that I ever used had some red test for some platform.
16:18:24AraqZevv: == is overloaded so it produces a nkSymChoice
16:18:47Araqmy guess is that the '==' export list is accurate
16:22:19ZevvBut it should only list the '==' from nativesocket, right? proc `==`(a, b: Port): bool. Now it links to other == implementations in system, options, winlean, etc
16:25:51Araqno, the code is:
16:25:54Araqimport nativesockets, os, strutils, parseutils, times, sets, options
16:25:54Araqexport Port, `$`, `==`
16:26:09Araqit's not export nativesockets.`==`
16:26:24ZevvAh I see, thanks for pointing that out
16:47:22thumbs_scumHappy Same Week everybody!
17:03:30FromGitter<arnetheduck> packagers (and users of certain distros) run on an entirely different set of hardware (when was the last time anyone ran test suite on `ppc64` and reported back results? it's an "officially supported platform"), software (c compiler, glibc and all other nim dependencies) etc than travis - of course the test suite is valuable outside of core development
17:07:47*Ven` joined #nim
17:10:05Araqarnetheduck: don't avoid the question. What do you do when tests are red/broken?
17:13:11*Ven` quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
17:24:48*kapil____ joined #nim
17:26:43shashlick@dom96 - sorry about the choosenim pr - will try and finish it soon
17:30:56*nsf joined #nim
17:51:54*narimiran joined #nim
17:56:39dom96We should probably remove ppc64 from our officially supported list
17:56:48dom96we used to run tests on it for every commit but no longer do
17:57:40stefantalpalarusegfault while compiling aporia-0.4.2 with nim-0.19.2: https://gist.github.com/stefantalpalaru/9d57b57e984a734edb9f947fcce29f37
17:58:06stefantalpalaruglib2 is the latest from https://github.com/nim-lang/gtk2
17:58:37narimirani haven't been online this afternoon. now i see in the irc logs some mesages from stefantalpalaru that something (testament) is missing from 0.19.2, and leorize has found https://github.com/nim-lang/Nim/commit/00f84d3d22cdeed7b4a2060e6307e3a742729055 which should have been backported.
18:00:04*OrganicAnywhere joined #nim
18:00:56narimiranshould i backport it now, if there will be 0.19.4 one day?
18:01:17*rect0x51 joined #nim
18:02:29shashlickwhile @kaushalmodi is away, i need some clear direction on the goals of nightlies
18:05:07rect0x51hey everyone, for bux fixes: do I use devel branch directly or a different branch?
18:05:33shashlickpresume you have your own fork
18:05:33OrganicAnywhereI tried to download choosenim-0.3.2_windows_i386.exe from dom's github and it got stopped after my computer detected a trojan. Does anyone know what happened?
18:05:52shashlickcreate a separate branch - that way, while the PR is getting reviewed/approved, you can continue to work on other stuff
18:06:22narimiranrect0x51: what shashlick said: pull devel, create a branch, make bugfixes, submit PR
18:06:37narimiranOrganicAnywhere: false positive, probably
18:09:09rect0x51In documentation I read I should rebase my branch. What really confuses me is... is you rebase a branch to devel, doesn't that branch become devel? Sorry if it's stupid question, I am new to git/github.
18:10:04rect0x51I probably worry more than I should, but I really want to understand how it works and make proper PRs.
18:10:15narimiranrect0x51: 1. fork nim-lang/Nim, so now you have your own rect0x51/Nim repo
18:10:36narimiran2. clone your repo, so now you have local copy of it
18:11:09OrganicAnywhereI googled the file name, "Zpevdo.A" and others have received this file when downloading other, non-nim related stuff.
18:11:29narimiran3. (if not already on `devel` branch: `git checkout devel`)
18:11:52stefantalpalaruthat Aporia segfault doesn't appear with Nim HEAD
18:11:59narimiran4. `git checkout -b myFixBranch`
18:12:02dom96OrganicAnywhere: It's not a virus. Anti-viruses look for embedded binaries and flag it up as a virus
18:12:14rect0x51Ok, let me describe what I've done so far. I forked Nim, cloned my fork (which adds a remote to my fork implicitly), add the upstream remote. Now I want to start working on a bugfix, what do I do? Also when/how often (if at all) should I rebase my local branch to the upstream devel to keep up to date?
18:12:22*Cthalupa quit (Ping timeout: 272 seconds)
18:12:44narimiran5. do your fixes, commit the changes, push it to your own repo — once you do, you'll see "make a pull request" on github
18:13:33*Cthalupa joined #nim
18:13:46rect0x51narimiran: basically what I am struggling with is how to "keep my fork up to date with upstream", but I am not sure if that's even needed.
18:13:46narimiranrect0x51: you did all the right steps. now you can start working on bugfixes and stop worrying :)
18:14:14dom96stefantalpalaru: does it happen with 0.19.0?
18:15:42narimiranrect0x51: you'll be fine without updating most (>95%) of the time. before starting a new bugfix, pull the latest devel and branch from it. rinse and repeat
18:16:04OrganicAnywheredom96, what is it then? I'm inclined to think it's a false positive but it's detecting a file that is not the file I'm downloading and others hav detected it as well.
18:16:24rect0x51narimiran: awesome, that's what I wanted to know, thanks!
18:16:41narimiranrect0x51: no problem, now i'm waiting for your PRs :)
18:17:01dom96OrganicAnywhere: submit it to virustotal, that should show you how other anti viruses are detecting it
18:17:22dom96if it makes you uncomfortable just don't use it and install Nim the traditional way
18:17:35rect0x51narimiran: you will have to wait a bit, I want to go through every Nim doc first (manual, GC, internals etc), then I have to start understanding the codebase somehow (newbie here)
18:17:38OrganicAnywhereAlright, thank you :)
18:18:19rect0x51but I am really determined to work on Nim, and as a uni student I have some spare time, haha
18:19:01dom96rect0x51: start by fixing bugs in the stdlib and/or bugs marked "Easy"
18:19:26rect0x51@dom96: sure, don't worry, I don't plan to start big
18:19:35dom96By far the best way to fix bugs is to fix the bugs that affect you, so write some software in Nim and you'll surely run into something that can be improved
18:20:17rect0x51aha, sounds like good advice. now i have to think of some interesting project (having serious trouble with that in general)
18:22:02dom96https://github.com/nim-lang/needed-libraries/issues ;)
18:22:51rect0x51ohh hadn't came across this page, excellent!
18:24:35*OrganicAnywhere quit (Quit: Leaving)
18:24:37dom96So awesome https://twitter.com/wilfredspringer/status/1079842955005972481
18:26:07shashlick@dom96: i'm working on nightlies right now, i'll update the choosenim PR later today if I get a chance
18:26:24shashlickbut am still a bit unclear if you wanted to remove the github fallback
18:27:10stefantalpalarudom96: that Aporia segfault does not appear with nim-0.19.0
18:29:17rect0x51I think I'd love to work on tooling (linting etc). What is Nim's biggest need regarding tooling atm?
18:29:51shashlicki'd think getting nimlsp working in more editors would be awesome
18:29:59*kidandcat[m] left #nim ("User left")
18:30:07shashlickalso work on nimble and choosenim will also be appreciated
18:31:15narimiranrect0x51: 0.19.2 has lots of nimpretty improvements, but there are still some open issues: https://github.com/nim-lang/Nim/issues?q=is%3Aopen+is%3Aissue+label%3Animpretty
18:31:52narimiranbut nimble and choosenim would be great too, i agree
18:33:54rect0x51nimsuggest basically gives you more hints than nimcheck?
18:39:38rect0x51ah, it provides autocompletion and stuff too, nvm
18:40:18rect0x51I've been using vim ALE which uses nim check, but I guess a solution like nimsuggest is more "Nim specialized".
18:41:21shashlickCheck out nimlsp which leverages nimsuggest and provides a standard language server instead of the specialized nimsuggest protocol
18:41:30shashlickWorks with vim apparently
18:44:02*thomasross quit (Read error: Connection reset by peer)
18:44:11*thomasross_ joined #nim
18:44:11*thomasross_ is now known as thomasross
18:48:07FromGitter<arnetheduck> Araq - you either fix it or manage expectations - the point I'm making is that the test suite is useful outside of development - making sure it works well is a good investment - it gives people that care about making nim work an opportunity to know something is wrong and eventually fix it
18:49:07rect0x51nimlsp surely is vital. irrelevant... but I found some weird behaviour while using nimgrep, though I'm skeptical about how useful nimgrep is (grep does the trick just fine)
18:49:43Zevvrect0x51: If you did not start from latest devel, of you have a devel-based branch lying around for some time, you can always fetch latest devel and rebase on that. Basically, this rewrites your local history and replays your commits on top of latest devel.
18:50:29FromGitter<arnetheduck> managing expectations is a service to the people using nim - if you call something "officially supported" for example, make sure there's a clear definition of what that entails, so I can take that information with me when I have to make a call about where I can apply nim
18:50:36rect0x51Zevv: yes, I figured that out after hours of reading about git. thanks anyway.
18:52:02rect0x51Zevv: it seems I should rebase before starting and after finishing, right?
18:52:14narimiranrect0x51: just before starting
18:52:42narimiranand only if after finishing there is some merge conflict, you can do rebase. otherwise, you'll be fine.
18:52:59rect0x51ok I think I get it
18:55:55shashlickI use nimgrep all the time now
18:57:21rect0x51shashlick: can you describe when/how it is useful? I tried to find the location of a string literal and it didn't work.
18:58:18Zevvit knows how to work with Nim's case and _ agnostic identifiers
18:58:30Zevvyou can grep for foo_bar and match fooBar
18:58:31shashlickrect0x51: this is how I use it - `alias ng="nimgrep --recursive --ext:$1 $2 $3 $4 $5 $6 $7 $8 $9"`
18:58:50shashlicknow I can `ng nim abc` and it searches only nim files for the string I search for
18:58:58rect0x51Zevv: yeah I read about that, but it seems non-robust, is what I am saying.
18:59:02shashlicki'm usually only looking in nim or h files
19:03:23rect0x51ok let me put that in context. I did nimgrep --recursive "'case' objects" and it returned 0 matches, while grep -RI "'case' objects" returns a match.
19:04:06rect0x51Am I missing something?
19:04:49*zachk joined #nim
19:04:49Zevvhehe funny
19:04:56Zevvsomething with the quotes I guess
19:05:18rect0x51maybe, but probably not, I tried many thing, some of them didn't have quotes
19:05:23shashlickyou need to escape quotes and stuff
19:05:59rect0x51same thing with "iterator does not work"
19:06:24rect0x51nimgrep --recursive "iterator does not work" 0 matches
19:07:09Zevv"case..objects" matches, but "case. objects" does not
19:07:48*zachk quit (Changing host)
19:07:48*zachk joined #nim
19:08:00Zevvescape the space: "'case'\ objects"
19:08:05Zevvnot intuitive no
19:08:11rect0x51ohhh ok that explains a lot
19:08:46rect0x51ok I have a bug, not sure if it's my terminal or whatever but...
19:09:04rect0x51it's really weird
19:12:03rect0x51https://asciinema.org/a/N1uu1xazY1cpvZp1rryUKEkyy
19:12:40rect0x51:S
19:23:05dom96You're grepping through binaries
19:23:19FromGitter<arnetheduck> dom96 - instead of removing, an option is to rephrase the section into something like: "we regularly run tests on platforms x. we make an effort to support y as secondary platoforms. z generally work and we encourage contributions"... then tie that to a release policy where tests must be green on primary platforms and an rc period can be used to collect test results on secondary ones - and make an informed choice
19:23:19FromGitter... about whether to proceed with the release or not.. also, if there are more frequent releases, the "cost" of `devel` breaking / not having green tests goes down, because there's less incentive to actually use it
19:24:13rect0x51@dom96: I see... is there an option to ignore binaries (like grep -I)?
19:27:53Araq--ext:nim
19:28:30Araqeverything is binary, it's a digital computer, the filtering is done by file extensions
19:28:55rect0x51Araq: :D
19:32:02*dddddd quit (Read error: Connection reset by peer)
19:32:24rect0x51nimble install nimlsp doesn't found the package
19:32:49rect0x51doesn't find*
19:33:33narimiranrect0x51: nimble install https://github.com/PMunch/nimlsp
19:33:52rect0x51narimiran: ty
19:37:42FromGitter<arnetheduck> in part, this is also why at status, we prefer to fork the parts of stdlib that we use extensively.. this has several advantages: ⏎ ⏎ 1) there is less PR churn between us and upstream which benefits efficiency of both camps - we can develop the code to a higher quality before suggesting its inclusion in mainline - we free up the time of Araq and Arne and whoever else is reviewing ⏎ 2) we avoid depending on
19:37:42FromGitter... nim (languague) releases to advance features which can be achieved in libraries - people wanting to use our stuff (including us) can do so earlier without having to follow `devel` ⏎ 3) we don't have to consider breaking the world as much - people can pin specific versions of the package in their dep list which lowers the ... [https://gitter.im/nim-lang/Nim?at=5c2bc186ab910e7d3a081086]
19:39:42Araqarnetheduck: I agree with you completely. but I don't understand why you bring it up?
19:41:09FromGitter<arnetheduck> because I emphasize with stefantalpalaru about that it's vital for 0.19.2 to come with a good option for running the test suite, and to show examples of where that's useful
19:42:06FromGitter<arnetheduck> also to show the reasoning that goes into choosing different development styles - devel vs releases, point releases etc - how we think about things as consumers of the language
19:42:47narimiranspeaking of which: https://irclogs.nim-lang.org/01-01-2019.html#17:58:37
19:42:58Araqthat we broke testament is unfortunate but it doesn't warrant any immediate action, IMO
19:43:35Araqespecially since I don't understand why packagers want to run the tests. The way I see it there are only two options.
19:43:37*kapil____ quit (Quit: Connection closed for inactivity)
19:44:11Araqa) you don't update the package. But then the *real* users of Nim might want to upgrade anyway because they want to test *their* own code with 0.19.2 and don't care about the test suite.
19:44:31Araqb) you do update the package anyway. But then, why run Nim's tests?
19:46:40Araqand maybe c) you wanna run the tests to see what it's broken on your platform in order to decide whether to update the package or not. But then, running the tests still works afaik, only 'koch tests' is broken
19:48:32Araqand I don't know why this decision should be taken away from the user anyway.
19:49:34shashlickAraq: winrelease is breaking on nightly
19:50:10shashlickhttps://api.travis-ci.org/v3/job/474151232/log.txt
19:51:04*nsf quit (Quit: WeeChat 2.3)
19:51:38FromGitter<arnetheduck> well, the whole point with distros is that they do the work of arriving at a set of packages that work well together, while at the same time catering to different audiences. for example, debian enables `pie` on all their executables as a security feature - in `nlvm` that doesn't work because it's set to generate `non-pie` code.. that's something I wouldn't know, unless someone ran the test suite on debian - it's
19:51:38FromGitter... pretty subtle. then you have the economy of the distro: the cost of verifying whether a package and version will work depends on how complicated it is to automate - so the chances of someone wanting to do that work depend on how willing upstream is to keep certain things stable / predictable - such as how to match ni ... [https://gitter.im/nim-lang/Nim?at=5c2bc4c937975e7ca95acc43]
19:54:38FromGitter<arnetheduck> so indeed, yeah, c) - some distros cater to users that appreciate that distros do this testing for them - it's a feature that its "taken away from the user".. others let users in to the process - like gentoo which has the option to run the test suite as part of the upgrade (in a uniform way, across packages)
19:58:10FromGitter<arnetheduck> what that does is tell them "will this release of nim work well with the particular mix of stuff I have on my system already?" - for distros like fedora or debian, it's fine that the test run is done once by the distro packager - the user has no great advantage once the packager has run the tests, to rerun them..
19:59:07Araqfair enough I guess. It's linux, after all.
19:59:56rect0x51I am using the devel build and nimble instal <link to nimlsp> fails with "nimlsppkg/Nim/nimsuggest/nimsuggest.nim(13, 10) Error: nimcore MUST be defined for Nim's core tooling"
20:00:22shashlickAraq: I don't see where web/upload/download is created - how is winrelease typically called?
20:00:42Araqtools/winrelease
20:01:26rect0x51do I have to tweak something in Nim/config/nim.cfg to make it install?
20:01:53shashlickwinrelease imports koch and koch doesn't create web/upload/download either
20:02:22Araqrect0x51: compile with -d:nimcore as a workaround
20:02:37Araqshashlick: probably I simply have these dirs on my machine
20:02:49shashlickam adding fixes to travis for now, will improve this once it works
20:04:03FromGitter<arnetheduck> that said, for something like nim that isn't a pure dependency for end-user features (unlike gcc or libc where users generally don't care about the exact version, as long as it works), the value of being in a distro isn't that high - specially if there's a simple way to set up a local isolated environment (aka choosenim)
20:05:31FromGitter<arnetheduck> I suspect for example very few people consume `rust` or `node` through their package manager, simply because the experience of an isolated install via `rustup` or `npm` is so good
20:06:40Araqand yeah, just to make this clear. Unfortunately 0.19.2 was built manually because nightlies still wasn't ready and I delayed this release for too long already
20:07:34shashlickAraq: so nightlies is only building and packaging, it does no tests
20:08:02Araqit needs to do 'koch testinstall' but there are also better options
20:08:06shashlicki presume the testing still relies on nim-lang/nim CI regardless of devel or version-0-19 branch
20:08:32Araqlike really downloading the zips, building them, running the tests with an almost empty PATH
20:17:32*vlad1777d quit (Ping timeout: 250 seconds)
20:18:49stefantalpalaruPlease reopen this issue, since it applies to 0.19.2: https://github.com/nim-lang/Nim/issues/9889
20:20:38stefantalpalaruMight want to also add a test case for it.
20:29:31narimiranstefantalpalaru: there are plenty of fixes which are not in 0.19.2
20:30:47stefantalpalaruFascinating, but this one is for a newly introduced bug. See the linked issue.
20:32:53ZevvWhen addind examples to stdlib documentation, are runnableExamples prefered over 'normal' examples?
20:33:23narimiranZevv: yes
20:34:21ZevvOk. I find the 'doAssert' notation a bit noisy, though, but I'll get used to that
20:35:07FromGitter<arnetheduck> @Araq what's the story with gcv2 btw? why is it a dead end?
20:41:16*narimiran quit (Quit: Leaving)
20:43:14*vlad1777d joined #nim
20:45:31Araqnot a deadend, but also not something I intend to work on
20:47:15*stefantalpalaru left #nim (#nim)
20:53:57rayman22201offtopic but interesting for compiler optimization perhaps: https://www.ps.uni-saarland.de/~sdschn/LVC.html
20:58:32*aziz joined #nim
20:58:49*aziz quit (Client Quit)
20:59:38shashlickAraq: got windows working somewhat on nightlies but build times out
20:59:47shashlickhttps://travis-ci.org/nim-lang/nightlies/jobs/474171958
21:01:06*stefantalpalaru joined #nim
21:07:23*stefantalpalaru quit (Changing host)
21:07:23*stefantalpalaru joined #nim
21:09:48Araqrayman22201: interesting indeed
21:15:57Araqshashlick: I need to think about it
21:16:15Araqby its nature it's a long process
21:17:20shashlickDo we need to do docs?
21:17:31shashlickDidn't think we need that for packaging
21:20:05FromGitter<arnetheduck> btw Araq, any decisions on https://github.com/nim-lang/Nim/issues/9017? would be really nice to have the information about matching nim & nimble versions in the git repo, but if that's a problem, we'll implement some ugly workaround
21:23:26*krux02 joined #nim
21:32:06Araqshashlick: the zips include the *.html docs
21:32:07*dddddd joined #nim
21:33:16Araqarnetheduck: your ugly workaround will be as ugly as git commit hashes in koch.nim
21:33:25Araq:-)
21:33:28*xet7 quit (Remote host closed the connection)
21:34:25FromGitter<arnetheduck> not really, because it's part of your release process .. you have to do the work anyway, it's a matter of writing it down in a reproducible way
21:34:27stefantalpalaruI vote for a Git submodule for Nimble.
21:34:57AraqI wrote it down, it's in koch.nim, we ship the latest stable Nimble release
21:35:47FromGitter<arnetheduck> I can't reproduce that. I'd have to know the exact time you ran the release process and get the latest stable release at that time.
21:36:57FromGitter<arnetheduck> the latest current stable release likely does not work with nim 0.17, agree?
21:37:37Araqagreed but why does it matter, the tar.xz contains nimble's source code in dist/
21:37:56*Lroy joined #nim
21:38:30Araqyou don't have to know anything the tarball contains the C sources, the Nim sources, the Nimble sources...
21:39:16FromGitter<arnetheduck> because git is more convenient than tarballs to build from - it allows testing patches, fixes, backports, bisecting, all kinds of things
21:39:44FromGitter<timotheecour> I agree with @arnetheduck
21:39:45shashlickAraq: I think winrelease builds nim and docs again
21:39:46Araqand if you don't trust the tarball there is little reason to trust anything, I could have inject a Thompson hack in 2010
21:40:07FromGitter<arnetheduck> it's not that I don't trust the tarball, it's that the feature set it offers is limited compared to git
21:40:38FromGitter<arnetheduck> the trust situation is similar:ish
21:41:10Araqhere is what you do:
21:41:22Araq- start with the tarball. Gives you a Nim.
21:41:35Araq- use github and the Nim to get newer Nims
21:42:07Araqeverything else is a never ending exercise in masochism.
21:42:37Araqwhere you dream up always more complex situations in how the build may fail
21:44:14FromGitter<arnetheduck> I don't have trouble getting *a* nim.. I'm having trouble knowing which nimble that goes with it - which nimble was used to produce the nim 0.19.2 release for example..
21:44:50Araq0.19.2 has a release date, nimble's tags have dates.
21:46:46FromGitter<arnetheduck> well yeah, that's the kind of ugly workaround we can do - it doesn't qualify as reproducible by most metrics though because there's an approximation in there
21:47:27FromGitter<arnetheduck> there's nothing in `nim` that binds the `nim` tag to a `nimble` tag beyond an external, casual relationship
21:49:28FromGitter<timotheecour> @arnetheduck it’s related to https://github.com/nim-lang/Nim/issues/9129
21:50:13Araqwould it help if the tar.xz had a generated file 'used_commits.txt' ?
21:50:29FromGitter<timotheecour> (either using a git submodule or adding a git hash in version control would be fine)
21:50:40FromGitter<arnetheduck> for example, if someone hijacks the nimble repo, it's easy to subvert (like happened recently with a node module).. you can stick your head in the sand and pretend these things don't happen, but unfortunately, it's the kind of stuff we have to realistically deal with as it happens on a regular basis in the real world: https://discuss.status.im/t/wallet-dependency-compromised
21:52:59Araqgit commits in a git repo don't work if they are not automated. and if they are, why are they in the github repo?
21:53:10FromGitter<arnetheduck> in an ideal world, we don't want to touch the tarball at all - it's just another complication - if we can stay 100% git, it makes it more easy for us to solve our problem, to contribute to nim, etc - git already does the secure code management to a sufficient degree - with the tarball, we'd have to do hashes and all that stuff
21:53:44FromGitter<arnetheduck> ...by hand
21:53:58AraqI mean seriously, what are we discussing here? I screwed up the [backport] for some testament commit
21:54:18Araqand now I'm to be trusted with maintaining nimble commit hashes in koch.nim?
21:54:22Araqmanually?
21:54:26Araqseriously?!
21:55:41FromGitter<arnetheduck> well, that's why git submodules are easier in a way - in your `release.sh 0.19.2`, which does *all* the steps, you include a step that records what was built and save that to git.. no manual steps involved
21:56:06Araqas I said, the tar.xz can get a gitcommits.txt
21:56:11Araqthat would make sense.
21:57:54Araqor a tar.xz that doesn't strip the .git directories
21:58:44FromGitter<timotheecour> does this help? https://blog.abelotech.com/posts/how-download-github-tarball-using-curl-wget/ ⏎ ⏎ eg: https://github.com/<username>/<repository>/tarball/<version> ⏎ => allows getting tarball from specific hash [https://gitter.im/nim-lang/Nim?at=5c2be29492cf4d22422c2605]
22:00:47FromGitter<timotheecour> (and there’s also git archive, https://alvinalexander.com/git/git-export-project-archive-cvs-svn-export, but 1st link seems fin)
22:00:50FromGitter<arnetheduck> well, it's an improvement - we would then: download the tarball, parse gitcommits.txt to find the nimble hash, throw away the tarball, use git to checkout that nimble version. keep in mind that the assumption is that we already know which `nim` hash corresponds to the nim release
22:01:18AraqI don't know why you throw away the tarball
22:02:32FromGitter<arnetheduck> because it's not useful for the aforementioned reasons: it doesn't support patches and branches
22:05:28Araqah I finally got it, you have the same problem that I have with nimble
22:06:33Araqthrowing away the .git directory is always a bad idea.
22:08:45FromGitter<timotheecour> Ya for many reasons (eg if we want to temporarily edit a file , u don’t have git to help you see what you changed; it also makes certain things less efficient (instead of `git checkout`, you have to update entire source tree when changing version of a nimble package)
22:10:18FromGitter<arnetheduck> well, we have many problems with nimble - that's one of them - but regardless, the point is that we *can* do all these complicated workarounds, like not use `build_all.sh`to build *a* nim or not use `koch tools` to build the right nimble version.. but that makes the zero-to-hero workflow longer / more complicated, if you want to stay within the git world and not bother with complications like tarballs
22:14:20Araqthere is no "git world without tarballs". in other cases you have a bunch of C++ files and are supposed to already have a C++ compiler or you go and download a clang.exe to kick off the process
22:15:23stefantalpalaru'there is no "git world without tarballs".' - why not, when everything we work with is already in Git repositories?
22:15:43Araqnot to mention that you need to have a git.exe to start with. And you also have to trust it enough. Maybe that git.exe injects a Thompson hack...
22:16:09Araqor maybe the Linux OS you run changes how git.exe runs and injects a Thompson hack
22:21:40*Lroy quit (Quit: Leaving)
22:29:47*Ven`` quit (Ping timeout: 240 seconds)
22:31:08rect0x51I don't understand what nimlsp does compared to nimsuggest... isn't simsuggest already a LSP?
22:31:11FromGitter<arnetheduck> divide and conquer. `nimble` is our most pressing issue right now because it actually breaks our build, and we'd like to take it out of the equation.
22:36:56*stefanos82 quit (Remote host closed the connection)
22:42:45*vlad1777d quit (Ping timeout: 268 seconds)
22:47:50FromGitter<timotheecour> see https://forum.nim-lang.org/t/4517#28229 [help needed] nim version of: COMPARING PYTHAGOREAN TRIPLES IN C++, D, AND RUST
22:49:32FromGitter<arnetheduck> for example, here's how we do the CI: https://github.com/status-im/nimbus/blob/master/.travis.yml#L36 - check out a specific `nim` hash, build it using the documented workflow for building nim & tools, run our tests. we don't have to worry about gcc, git etc because they (can) come from a locked-down distro (for example docker). there are two moving parts in that setup really - `build_all.sh` that gets a random
22:49:33FromGitter... `csources` and `koch tools` which gets a random nimble... our options are: ⏎ a) upstream fixes this so the `nim` hash also covers `nimble` and `csources` in terms of reproducibility ⏎ b) we have to do a workaround - save good `nim`, `nimble` and `csources` hashes somewhere, replace `build_all.sh` and `koch tools` wit ... [https://gitter.im/nim-lang/Nim?at=5c2bee7cbabbc178b211eadb]
23:07:18shashlickAraq: will winrelease not work with Nim built from csources?
23:07:44shashlickCause it rebuilds Koch, Nim and docs
23:08:02shashlickFigured I'd call it directly after building csources but it fails
23:08:15*vlad1777d joined #nim
23:09:03Araqkoch boot -d:release is required
23:10:09Araqrect0x51: nimsuggest uses its own protocol because it predates the LSP
23:10:41*Vladar quit (Remote host closed the connection)
23:11:58rect0x51Araq: I still didn't manage to install nimlsp. The command you suggested seems to install dependancies only (-d option of nimble install), I tried running nimble install <github link to nimlsp> and after, but I get the same error about nimcore.
23:12:24Araq--define:nimcore
23:12:38Araqadd it to some config if the command line fails you
23:13:08rect0x51--define doesn't exist
23:15:29rect0x51(and -d is depsOnly) anywho, I don't know how to define nimcore, or what that even means...
23:16:08shashlickAraq: you don't need koch csource -d:release for windows right?
23:18:20Araqshashlick: true but then also ensure we don't have a build/ dir in the zip. would be confusing otherwise
23:18:37Araqarnetheduck: Will have a look at it tomorrow and come up with a solution
23:19:36shashlickokay i'll get the build working but then need your help to ensure what is built is what is expected
23:22:54*krux02 quit (Remote host closed the connection)
23:23:00rect0x51I guess I should give up and just build from source
23:24:58FromGitter<arnetheduck> much appreciated - and like I said, we can work around it if it's too much of a bother - a key reason we're doing all this is because we want to be able to provide "simple" instructions to build nimbus exactly like we do it in our daily development workflow , CI etc - this is very attractive to contributors and people we cooperate with that are not into nim yet - because we dogfood the process, we have less to worry
23:24:59FromGitter... about in terms of annoyances they'll encounter along the way
23:26:17FromGitter<arnetheduck> another q: found a mem leak in gc1 - worthwhile to report?
23:31:57*vlad1777d quit (Ping timeout: 244 seconds)
23:34:13Araqthere are no memory leaks known
23:34:57Araqand usually it only looks like a GC bug, usually it's a codegen or stdlib bug
23:36:37Araqbut there are no codegen bugs known either that can cause memory leaks or stdlib bugs
23:36:58Araqso definitely worthwhile to report
23:38:57FromGitter<arnetheduck> ok. it's a codegen bug I think - it generates a `copyStringRC1` where it shouldn't (at least I think it shouldn't, if I understand it right)
23:43:26FromGitter<kaushalmodi> shashlick: about nightlies, I think the goal is to just build Nim distributables on each platform, and test that archive on the same platform using "koch testinstall" (that was failing the last time I tried)
23:44:19FromGitter<kaushalmodi> The stretch goal, I think.. @Araq correct me if wrong, is to also build the nim binaries for each major OS flavor/bits combo and deploy those.
23:44:50rect0x51... nimlsp doesn't get build with nimble, after defining nimcore I get "Error: undeclared identifier: 'Channel". I think I overestimated the state of the project, it seems heavily WIP.
23:45:19Araqno, you're missing its config somehow, --threads:on --d:nimcore
23:46:26rect0x51but shouldn't it be easy to build (with just "nimble install")?
23:46:33Araqkaushalmodi: maybe, eventually.
23:46:34shashlickthanks @kaushalmodi
23:46:37shashlicki have some fixes
23:46:42shashlickbut i think there's still some ways to go
23:47:10Araqrect0x51: I don't know the state of the project and PMunch is not in #nim right now
23:47:25shashlick@kaushalmodi: looks like windows has passed for 0.19.2 - https://travis-ci.org/nim-lang/nightlies/builds/474220119
23:47:49rect0x51ok, I guess I'll have to talk with someone who actually works on it.
23:58:19shashlickAraq, kaushalmodi - nightlies build has passed