<< 11-01-2018 >>

---

00:09:07	*	alexday joined #nim
00:09:11	alexday	hey
00:09:32	FromGitter	<data-man> Does anyone use this in Windows? https://github.com/Maximus5/ConEmu
00:09:36	dom96	hi alexday :)
00:12:40	alexday	is there a nil type in nim? I am implementing a simple trie with nim. I have type NodeObj = object children: seq[ref Node], isTermal: false. but when I initialize a new Node which is type Node = ref NodeObj. how can I make sure that (in terms of C) children[i] = NULL ?
00:13:06	alexday	I can think of like setting an isEmpty property in the NodeObj .
00:13:17	dom96	you shouldn't be using double 'ref's
00:13:22	dom96	seq[Node] should be enough
00:13:36	alexday	ok
00:13:54	dom96	you can just write: type Node = ref object ....
00:14:17	alexday	noted.
00:14:18	dom96	and if you want to make something nil just assign nil to it
00:14:26	dom96	maybe I'm misunderstanding though
00:16:52	alexday	so if I assign nil say, I would have to have something like a fixed size of children, and then iterate and set null. I was thinking, since seq[Node] is dynamic, is there anyway to check if the children[0]'th node is not empty or NULL
00:17:15	alexday	without pre-assigning fixed set of child nodes
00:17:39	FromGitter	<brentp> @mratsim good to
00:17:42	FromGitter	<brentp> know
00:17:52	FromGitter	<brentp> Cigar:https://genome.sph.umich.edu/wiki/SAM#What_is_a_CIGAR.3F
00:19:45	*	gangstacat quit (Quit: Ĝis!)
00:20:58	dom96	alexday: check the length and then the contents
00:21:06	alexday	ok. yeah
00:21:15	alexday	that sounds cool. stupid me :D
00:23:26	*	gangstacat joined #nim
00:23:35	*	gokr quit (Ping timeout: 276 seconds)
00:27:33	GitDisc	<lotzz> Hello everyone
00:32:10	*	dddddd quit (Remote host closed the connection)
00:34:35	*	MJCaley joined #nim
00:45:45	FromGitter	<data-man> Hello!
01:01:19	*	kafke quit (Ping timeout: 248 seconds)
02:02:45	*	S1tiSchu joined #nim
02:03:28	*	MJCaley quit (Quit: MJCaley)
02:06:56	*	S1t1Schu quit (Ping timeout: 276 seconds)
02:10:28	*	zahary quit (Quit: Leaving.)
02:17:30	*	chemist69 quit (Ping timeout: 246 seconds)
02:19:14	*	MJCaley joined #nim
02:29:46	*	MJCaley quit (Quit: MJCaley)
02:32:01	*	chemist69 joined #nim
02:47:31	*	sz0_ joined #nim
02:49:55	*	sz0_ is now known as sz0
02:56:31	*	Cthalupa quit (Ping timeout: 248 seconds)
03:00:00	*	Cthalupa joined #nim
03:11:04	*	marenz_ joined #nim
03:15:11	*	marenz__ quit (Ping timeout: 276 seconds)
03:25:35	FromGitter	<saolof> does Nim distinguish between statements and expressions?
03:26:45	FromGitter	<RedBeard0531> yes, but most statements also have expression forms and there is a way to use multiple statements as an expression
03:27:10	FromGitter	<RedBeard0531> what are you trying to do?
03:27:37	FromGitter	<saolof> ah, I was just curious, I'm taking a look at the language and the syntax seemed python-like
03:29:29	*	vlad1777d quit (Ping timeout: 276 seconds)
03:29:30	FromGitter	<RedBeard0531> it is. but there are also things like https://nim-lang.org/docs/manual.html#statements-and-expressions-statement-list-expression and https://nim-lang.org/docs/manual.html#statements-and-expressions-case-expression that blur the line a little bit
03:56:23	*	endragor quit (Remote host closed the connection)
04:12:20	*	endragor joined #nim
04:19:35	shashlick	@data-man: I use ConEmu, pretty nice
04:23:49	FromGitter	<data-man> @shashlick: Which Windows version are you using?
04:24:06	shashlick	10
04:24:50	FromGitter	<data-man> With all updates? :)
04:26:22	shashlick	I hope 🙂 ... it is build 1607 Enterprise
04:27:13	FromGitter	<data-man> Can you check terminal module from my PR https://github.com/nim-lang/Nim/pull/6936 in the ConEmu? Please.
04:29:23	shashlick	sure
04:29:42	shashlick	what exactly do I need to do?
04:29:58	FromGitter	<data-man> ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5a56e846ba39a53f1affe3ca]
04:35:01	FromGitter	<data-man> What you see? :-)
04:35:51	shashlick	cannot open strformat?
04:36:51	FromGitter	<data-man> Oh! Isn't Nim devel?
04:37:36	shashlick	let me choosenim update #head
04:46:59	shashlick	https://imgur.com/a/M5MvE
04:47:11	shashlick	left is conemu in default color scheme, right is cmd.exe
04:47:25	*	arnetheduck quit (Remote host closed the connection)
04:48:42	shashlick	looks bad in Monokai colorscheme
04:50:05	FromGitter	<data-man> Big thanks! ⏎ I will try ConEmu on Windows 8 in the VirtualBox.
04:50:25	shashlick	no problem
04:50:59	*	arnetheduck joined #nim
04:52:20	*	arnetheduck quit (Remote host closed the connection)
04:53:51	*	arnetheduck joined #nim
04:54:57	FromGitter	<data-man> @shashlick: Are you using Far Manager?
04:55:20	*	Atomic2 quit (Ping timeout: 260 seconds)
04:56:09	shashlick	nope, just ConEmu
04:58:51	FromGitter	<data-man> Why? Far Manager so cool! And I'm one of the contributors. :-)
05:03:48	shashlick	I used to use norton commander some decades ago, now so used to explorer
05:04:08	shashlick	of course, with fzf and ff (my tool), I avoid explorer too mostly
05:04:33	shashlick	https://github.com/genotrance/ff
05:06:10	FromGitter	<data-man> Did you see my message to you about the Everything?
05:08:26	*	Elronnd quit (Quit: BRB)
05:09:17	*	Elronnd joined #nim
05:09:24	shashlick	ya i did try out everything but the fzf approach is much faster
05:09:46	shashlick	plus I can run any tool I want on any selection with shortcuts
05:10:08	shashlick	like I can do "ff T" and kill a process
05:11:06	shashlick	i've also bound it to a hotkey with autohotkey so now all I do is WIN-/ d c and I can open a command line window on the sub-directory I picked on my desktop
05:11:43	FromGitter	<data-man> Of course, its own utility is always better. :-)
05:12:35	shashlick	agreed, but it's worth a try, especially when it comes to coding - really speeds things up
05:13:12	FromGitter	<data-man> Remains to write NimFZF
05:13:19	shashlick	i even have a conemu shortcut to open a new cmd tab in a directory selected
05:14:07	shashlick	I've stayed away from terminal/ncurses stuff historically, kudos to you on your PR
05:14:09	shashlick	and the Far work too
05:15:35	*	kordanza joined #nim
05:16:10	FromGitter	<data-man> Oh, no problem! :-)
05:16:56	shashlick	where are you located?
05:19:50	FromGitter	<data-man> Can you login to gitter?
05:19:53	*	Elronnd quit (Quit: brb)
05:20:03	*	elronnd joined #nim
05:21:10	*	elronnd is now known as Elronnd
05:22:26	shashlick	how does that help?
05:24:18	FromGitter	<data-man> I can write to you privately.
05:28:42	*	kordanza quit (Quit: Leaving)
05:33:21	*	Atomic2 joined #nim
05:34:22	*	Atomic2 quit (Client Quit)
05:52:37	*	nsf joined #nim
06:11:07	*	arecaceae quit (Remote host closed the connection)
06:11:26	*	arecaceae joined #nim
06:11:45	*	marenz_ quit (Ping timeout: 260 seconds)
06:14:08	*	arnetheduck quit (Remote host closed the connection)
06:41:34	*	vlad1777d joined #nim
06:47:27	*	arnetheduck joined #nim
06:55:44	*	gokr joined #nim
07:56:50	*	yglukhov joined #nim
07:57:25	*	yglukhov quit (Read error: Connection reset by peer)
07:57:37	*	yglukhov joined #nim
08:07:50	*	PMunch joined #nim
08:10:13	FromGitter	<survivorm> @Yardanico you're using arch, aren't you?
08:10:54	FromGitter	<survivorm> How do you install nim/nimsuggest?
08:13:24	miran	choosenim
08:13:38	Araq	koch nimsuggest
08:25:42	*	floppydh joined #nim
08:25:58	*	claudiuinberlin joined #nim
08:30:31	FromGitter	<survivorm> i'm a looser... ⏎ no installations of nimsuggest seem to work properly on my PC
08:31:37	Araq	what editor are you using?
08:31:57	PMunch	survivorm, I'm using Manjaro (Arch based) and choosenim works fine for me :)
08:32:07	FromGitter	<survivorm> ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5a572107b48e8c3566c81ac7]
08:32:41	FromGitter	<survivorm> As you see, none given ANYTHING
08:32:49	FromGitter	<survivorm> not even one
08:32:59	Araq	check nimsuggest.log maybe
08:33:57	Araq	oh and all these commands require a file.nim argument
08:34:07	Araq	but don't tell you you need it :P
08:34:34	Araq	nimsuggest is mostly used over sockets and it used to be chatty
08:34:45	Araq	this caused all sort of protocol related problems
08:34:54	Araq	and so now it shuts up completely.
08:35:51	FromGitter	<survivorm> https://gist.github.com/survivorm/c9953b6e361cd7999a4bc337423ac8a0
08:36:24	Araq	cmd: sug, file: 36, dirtyFile: [11:0]
08:36:24	Araq	Error: invalid module name: ''
08:40:49	FromGitter	<survivorm> input file, @@34-36 ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5a572311290a1f456175b1d0]
08:44:02	FromGitter	<survivorm> btw, code is compiling and working, if that matters
08:44:33	FromGitter	<survivorm> i may provide full example, if needed
08:47:58	FromGitter	<Yardanico> @yglukhov gist works for me, I posted a gist raw link
08:48:26	Araq	survivorm: give it an absolute filename
08:48:57	FromGitter	<survivorm> in starting, or in sug/def ?
08:49:04	Araq	in sug/def
08:49:04	FromGitter	<survivorm> @Araq
08:49:14	FromGitter	<survivorm> ok, i'll try
08:51:12	FromGitter	<survivorm> gave me something
08:51:22	FromGitter	<survivorm> it's great
08:52:03	Araq	PRs welcome
08:52:58	FromGitter	<survivorm> in improving nimsuggest?
08:53:11	Araq	yes
08:54:10	Araq	I didn't consider these problems of importance as the tools feed nimsuggest the commands in the way it requires
08:54:13	FromGitter	<survivorm> And that may be the place i'll start digging. Even if only because of how much blood it costed myself
08:54:33	Araq	nimsuggest is not for humans :-)
08:54:40	FromGitter	<survivorm> i now
08:54:59	FromGitter	<survivorm> but it doesn't work in editors, too
08:55:19	Araq	it does for VSCode
08:56:13	FromGitter	<survivorm> I tested it in textadept, vim, qt creator and some more on my PC (with pluggins, of cource) and so far - to no avail
08:57:40	FromGitter	<survivorm> Maybe my before's install was broken. Maybe i've got unlucky. Maybe it's system running version was corrupt - don't now, All i now - it didn't, for me
08:58:14	FromGitter	<survivorm> So, testing it and making it work better sounds like a plan
08:59:54	FromGitter	<survivorm> But i think i need a little more maths on parsers/compilators for this task :) ⏎ Currently trying to dig Ullman/Aho THE THEORY OF PARSING, ⏎ TRANSLATION AND COMPILING
09:01:09	FromGitter	<survivorm> Good book, but unfortunately I'm not so good at maths to read and understand it for more than 2-4 pages a day :)
09:03:17	*	gmpreussner quit (Ping timeout: 255 seconds)
09:05:03	*	gmpreussner joined #nim
09:08:30	Araq	meh, forget about that book
09:10:17	PMunch	Hmm, slight annoyance with nimble. I have a main file and a support file in my project. I added both to "src" but it asked me to move the support file into a folder named the same as the project and the main file. Problem is that trying to compile the main file now gives me an error saying that it's output already exists and is a folder as it tries to compile it to the same name as the folder..
09:10:56	euantor	@PMunch I've complained about that a few times :P
09:11:16	euantor	I still find it unexpected that compiling from the root directory tries to output an executable in the src/ dir
09:11:28	euantor	THe solution is to pass the `-o` option to change the output filename
09:11:38	Araq	Nim supports --output, I think nimble does the same
09:13:16	Araq	Unix simply requires .exe extensions
09:15:04	Araq	or maybe some lovely double fuck like exeNameHere.elf
09:15:45	Araq	exegcc, execp, exemv, execlang, exenim -- what's there to dislike? you can easily delude yourself it adds "clarity" to your system
09:17:34	alexday	https://w.awalgarg.me/d/NiRQ3HLJu can someone please help me with this? I pasted the error as comment at the end of the code
09:18:18	Araq	for s in toSeq(value.items) ?
09:18:54	Araq	for s in value # sometimes Nim is like Python
09:19:12	alexday	ow ok
09:19:26	Araq	let index = int(s) - int('a')
09:19:26	Araq	# ?
09:19:34	Araq	how does this addressing work?
09:19:47	*	Vladar joined #nim
09:20:46	Araq	anyway, you need to use crit bit trees from the stdlib
09:20:56	Araq	they do what you attempt to do
09:20:59	Araq	I think.
09:21:52	Araq	or at least learn about setLen and sequence indexing, it doesn't work as in Perl
09:24:03	alexday	ok will look
09:43:13	alexday	that adressing is just like the difference between character and 'a'. so like z - a is 25
09:43:22	alexday	and that's the array index
09:43:54	Araq	yeah I know but your seqs are empty
09:44:09	Araq	and you should have used an 'array' instead
09:51:44	*	sendell joined #nim
09:53:35	FromGitter	<Yardanico> @yglukhov maybe you checked gist yesterday, it was made with my old nickname so link was invalid
09:53:51	yglukhov	Yardanico: yeah, most likely
09:54:30	yglukhov	anyways, inlining the crash log into the issue looks like a more bullet-proof approach
09:56:31	*	Atomic2 joined #nim
09:56:50	*	c0ntribut0r joined #nim
09:57:03	*	c0ntribut0r quit (Client Quit)
09:57:39	*	c0ntribut0r joined #nim
09:57:55	*	Atomic2 quit (Remote host closed the connection)
09:58:19	*	gokr quit (Read error: Connection reset by peer)
10:09:05	*	Yardanico joined #nim
10:10:44	FromGitter	<Yardanico> well it's pretty big
10:11:09	FromGitter	<Yardanico> but OK, I'll inline crash logs in issues directly, I know that's better for the future :)
10:19:49	Araq	do we have a simple and fast zip algorithm written in pure Nim?
10:25:48	Yardanico	Araq, I don't think so :(
10:26:36	Yardanico	it seems that everyone uses libzip
10:27:02	FromGitter	<data-man> @Araq: https://github.com/jangko/nimz ? ⏎ I not testing :-)
10:27:08	Yardanico	oh wow!
10:27:15	Yardanico	sorry, I'm wrong then
10:27:21	Yardanico	but we don't know if it's simple or fast :)
10:28:00	Yardanico	data-man: how did you find it? :D
10:28:20	FromGitter	<data-man> It's my big secret :-)
10:28:33	FromGitter	<codenoid> he is data man
10:28:35	Yardanico	well I'm good at finding nim projects too
10:28:40	Yardanico	everyone knows that :P
10:34:16	FromGitter	<data-man> @Yardanico: You're amazing, no one is arguing that. :-)
10:38:09	FromGitter	<Yardanico> @GULPF hello, maybe you're around, what would be the best way to select random date if I have start and end of range? ⏎ currently I use "let randdate = fromUnix(int64(float(min) + rand(1.0) * float(max.toUnix - min.toUnix)))"
10:39:50	FromGitter	<Yardanico> min and max are both of "Time" type
10:39:52	livcd	hmm i am not able to install c2nim. Should I try to install nim 0.7.13 first ?
10:40:12	Araq	livcd, can you fix c2nim please?
10:40:18	Araq	I can guide you
10:40:26	Araq	the change should be one or two lines
10:40:35	livcd	ok lets try
10:41:15	Araq	fork it from https://github.com/nim-lang/c2nim
10:41:45	livcd	ok
10:42:02	Araq	c2nim.nim
10:42:08	Araq	the lines like
10:42:09	Araq	renderModule(tree, outfile, outfile)
10:42:12	*	Tanger quit (Ping timeout: 268 seconds)
10:42:14	Araq	should become
10:42:30	Araq	when compiles(renderModule(tree, outfile, outfile)):
10:42:38	Araq	renderModule(tree, outfile, outfile)
10:42:42	Araq	else:
10:42:47	Araq	renderModule(tree, outfile)
10:43:07	Araq	it's used in two places apparently
10:43:19	Araq	then c2nim should compile again against an older Nim
10:44:29	*	BitPuffin joined #nim
10:45:53	*	xet7 joined #nim
10:46:38	livcd	and the line 97? : renderModule(m, outfile, outfile)
10:48:47	Araq	similar.
10:49:51	livcd	also I have not found how to install this local version with nimble
10:50:12	Araq	"nimble develop" in c2nim's directory
10:50:28	Araq	you don't have to install it
10:50:44	Araq	just ensure it compiles with your Nim version and maybe run its testsuite
10:51:05	Araq	and then use your fork to create a PR
10:51:20	Araq	this step will run the tests anyway
10:51:39	Araq	it's fool proof muhahaha
10:51:47	livcd	really? but it won't be in the output right ?
10:51:58	Araq	what output?
10:55:10	*	Tanger joined #nim
10:56:54	Yardanico	wait, https://github.com/nim-lang/Nim/issues/2090 is this really implemented? AFAIK csources are still 1-core only :)
10:57:02	livcd	that it ran tests
10:57:32	Yardanico	ah, it's not for nim's csources
10:57:53	*	xet7 quit (Ping timeout: 265 seconds)
11:14:32	*	gangstacat quit (Quit: Ĝis!)
11:15:33	*	xet7 joined #nim
11:22:48	FromGitter	<GULPF> @Yardanico `let randdate = min + rand(0..(max.toUnix.int - min.toUnix.int)).seconds` is better since casting `Time` to `float` will break soon.
11:23:02	Yardanico	ok, thank you!
11:24:51	*	vlad1777d quit (Ping timeout: 240 seconds)
11:28:29	Yardanico	Araq, is this supposed to work? https://gist.github.com/Yardanico/cd3fb7a093a177da255335cb9d90c632
11:29:17	Araq	not sure, I think so?
11:29:39	Yardanico	well I mean fmt at compile time works, but this one fails (probably because of a template)
11:29:48	*	gangstacat joined #nim
11:31:01	Araq	Yardanico, make the template .dirty and it works :-)
11:31:10	Yardanico	ah, ok :)
11:31:19	Araq	otherwise the hygenic symbol binding rules bite you
11:43:07	Yardanico	oh wait, fmt actually takes raw strings? because fmt"{a[\"a\"]" doesn't compile (a is a table)
11:43:14	Yardanico	of course I can still use triple quotes :)
11:46:20	Araq	fmt is nothing special
11:46:33	Araq	fmt"" is an extended raw string literal
11:46:38	Araq	fmt "" is not
11:46:43	miran	Yardanico: aren't you missing } ?
11:47:07	Yardanico	miran, well I do, but that doesn't really matter here
11:47:23	Yardanico	Araq, ah, ok, thanks again
11:48:05	*	MyMind joined #nim
11:49:53	*	Sembei quit (Ping timeout: 248 seconds)
11:50:09	*	zahary joined #nim
12:02:12	*	Snircle quit (Quit: Textual IRC Client: www.textualapp.com)
12:07:34	miran	Yardanico: have you received your hacktoberfest t-shirt?
12:08:05	Yardanico	miran, yep
12:08:20	miran	nice. i didn't :(
12:08:24	Yardanico	:(
12:11:07	miran	when did it come?
12:12:52	Yardanico	https://i.imgur.com/ay5kYYy.png
12:14:57	miran	pfffft, before new year, and you were telling us sad stories about russia being last in the line.... :)
12:15:01	Yardanico	:P
12:24:18	Yardanico	btw, @alehander42 you rock! py2nim is getting advanced features much faster than I could've imagined: https://github.com/metacraft-labs/py2nim/issues/4
12:24:26	*	LyndsySimon quit (Read error: Connection reset by peer)
12:24:42	Yardanico	stuff like hasattr/generics/isistance
12:24:47	*	LyndsySimon joined #nim
12:27:13	*	literal quit (Ping timeout: 256 seconds)
12:27:47	*	literal joined #nim
12:29:30	*	zahary quit (Read error: Connection reset by peer)
12:37:41	alexday	:P check in the json file for the nim repos
12:37:46	alexday	:P check in the json file for the nim repos
12:38:26	alexday	wut. I replied to a super old message. twice.
12:38:27	alexday	my bad
12:48:32	*	zahary joined #nim
13:01:45	miran	"Also in the near future, -d:safety will be available to use along with -d:release for performance and safety (see: #2809)"
13:02:00	miran	is this implemented? or any news about it?
13:02:43	*	Snircle joined #nim
13:08:39	*	ShalokShalom joined #nim
13:10:51	*	zahary quit (Ping timeout: 256 seconds)
13:11:41	*	ShalokShalom quit (Remote host closed the connection)
13:14:59	*	zahary joined #nim
13:15:14	Yardanico	not yet TM
13:15:26	*	marenz_ joined #nim
13:15:37	Yardanico	you can make it by yourself since -d:release is defined in config/nim.cfg in main nim repo
13:24:01	dom96	make it yourself and make a PR to add it for everyone :)
13:27:58	*	zahary quit (Read error: Connection reset by peer)
13:28:11	*	zahary joined #nim
13:34:39	FromGitter	<survivorm> @PMunch do you plan for xrc in wxnim? I think it would be great
13:35:16	FromGitter	<survivorm> and this will add more compatibility with the external tools, too
13:36:08	PMunch	xrc?
13:36:19	FromGitter	<survivorm> https://wiki.wxpython.org/XRCTutorial
13:36:39	FromGitter	<survivorm> Generated by many GUI builders
13:36:52	PMunch	Huh, not sure how I've never heard of that :P
13:37:06	FromGitter	<survivorm> The idea is to separate the design from code
13:38:17	FromGitter	<survivorm> http://wxglade.sourceforge.net/, https://github.com/wxFormBuilder/wxFormBuilder generate it as a result. Glade too, of cource :)
13:38:20	PMunch	Yeah, similar to how Gtk has XML loading
13:38:54	PMunch	It's a good idea, although a bit weird since you can change it without changing code
13:38:55	FromGitter	<survivorm> It's gtk-related thing, you know :)
13:38:58	PMunch	For better or worse :P
13:39:14	Araq	form designers lead to non-generic code
13:39:41	PMunch	What do you mean?
13:39:42	Araq	better is a DSL that describes the forms IMO
13:40:02	FromGitter	<survivorm> Compatibility? Reuse?
13:40:25	PMunch	Yeah, that's why I created the GenUI DSL for various graphical toolkits :)
13:40:40	Araq	I mean with Delphi to create 5 check boxes I'll have to insert 5 check boxes on my formular
13:40:56	FromGitter	<survivorm> A-a-and?
13:41:09	PMunch	Oh you want to be able to do for i in 0..5: checkbox() ?
13:41:19	PMunch	Or something similar
13:41:43	Araq	but these likely come from some data model that is not captured at all
13:41:47	alexday	what does TObject mean?
13:42:36	Araq	and so I have code that sets these 5 checkboxes according to the underlying object my UI represents and also to read it back
13:42:42	PMunch	alexday, Type Object as opposed to PObject which is Pointer Object (or rather a ref in Nim). It's deprecated style though
13:43:19	PMunch	I feel there is a good way to make UIs, just that no-one have discovered it yet..
13:43:25	*	zahary quit (Quit: Leaving.)
13:43:26	Araq	PMunch, kind of yes, for i in 0..4: checkbox()
13:43:45	PMunch	Oh yeah, I always forget that .. is inclusive :P
13:44:11	FromGitter	<survivorm> @Araq I'll look into it, but may be, yes. But it may mean the wrong domain. There are grid and table controls, and may be they ARE not limited
13:44:23	FromGitter	<survivorm> maybe not
13:44:24	alexday	oh I see. that's why I can't see it in the docs? or maybe I am blind
13:44:28	PMunch	I still hope to be able to complete my type-based UI system. But I'm still having trouble with it..
13:44:34	Araq	oh it's not hard the problem is that you have 2 representations of the same thing and so you struggle with the consistency problems
13:45:42	Araq	immediate mode UIs don't have these problems, you have your program state not in a messy polymorphic non-generic tree structure
13:45:57	Araq	and everything becomes much simpler.
13:46:02	FromGitter	<survivorm> But idea is, for the menus and regular staff, this may be life-saver. And for something more... I'm thinking about something about Jinja2 on top of xrc
13:46:19	PMunch	alexday, well TObject is still used I think when ref is the default type. So you would have Something = ref TSomething if you use the reference by default or PSomething = ref Something if non-reference version is default
13:46:21	PMunch	IIRC
13:46:34	Araq	PMunch, it's called system.RootObj now
13:46:52	PMunch	Oh yeah, I was talking about the general naming convention
13:47:05	PMunch	TObject -> RootObj?
13:47:30	Araq	PObject -> RootRef
13:47:33	Araq	yes.
13:47:37	PMunch	Right
13:47:49	Araq	I still love immediate mode :P
13:48:09	PMunch	Immediate mode is nice in it's simplicity. But not very resource friendly..
13:48:35	PMunch	Updating the entire UI at 30FPS seems a bit of a waste if nothing actually changes..
13:48:47	Araq	there is no other way.
13:49:49	dom96	PMunch: Note: the common type should have no suffix, so in most cases PObject -> Object
13:49:57	FromGitter	<survivorm> Who was talking about resource-saving and economy :)
13:50:42	PMunch	dom96, yeah I know. That's what I was talking about before Araq mentioned RootObj which was a more direct answer to alexdays question :)
13:51:15	Araq	where are the benchmarks? seems unlikely in the presence of all these software layers that the list of rectangles to be updated makes it to the hardware.
13:51:35	Araq	I bet the HW redraws the whole screen instead :P
13:52:19	FromGitter	<survivorm> No bets as i have no idea how do they work ^P
13:52:22	Araq	it's a stupid idea anyway, what if you need to update the full window at 30FPS anyway (watching a video anybody?)
13:52:53	Araq	are then tell your users "you're fucked, can only update a smart part of the screen because of efficiency"
13:53:24	Araq	the worst case is quite common, deal with it.
13:57:42	Araq	the only problem with immediate UIs is where to put the state that is UI specific (for animations etc.)
13:58:45	*	dddddd joined #nim
13:58:53	Araq	the other problem is that most APIs are for retained mode
14:00:44	*	chemist69 quit (Ping timeout: 255 seconds)
14:01:58	*	arecaceae quit (Remote host closed the connection)
14:02:17	*	arecaceae joined #nim
14:03:08	*	chemist69 joined #nim
14:03:18	FromGitter	<alehander42> thanks @Yardanico , all of those are also very alpha yet, hasattr / isinstance mostly work for 1,2 situations in libraries I work on translating: basically a very early form of flow-based type checking (but eventually they'll improve haha)
14:04:31	alexday	PMunch thanks
14:04:38	FromGitter	<alehander42> btw why "" escapes " in strformat, instead of \"
14:09:56	Yardanico	fmt "look, I can use \" escape now!"
14:10:07	Yardanico	fmt("look, I can use \" escape now!") that's more clear ? :)
14:10:15	Yardanico	I mean that there are different rules]
14:10:35	Yardanico	<Araq> fmt"" is an extended raw string literal
14:10:38	Yardanico	<Araq> fmt "" is not
14:12:29	FromGitter	<GULPF> to bad the syntax highlight in vs code doesn't understand that
14:13:45	FromGitter	<GULPF> `fmt("\"")` and `fmt"\""` is highlighted the same, showing `\"` in a different color indicating it's a escape sequence
14:17:42	dom96	I wonder if we should make a bunch of syntax highlighter tests and shame everyone that doesn't abide.
14:19:21	FromGitter	<alehander42> argh so I have to use fmt<space> *sometimes*
14:20:15	Araq	https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/oopsla065-burckhardt.pdf
14:20:38	Araq	alehander42 same like you "sometimes" have to use r"" vs ""
14:21:05	FromGitter	<alehander42> but not for each \n and ", which are very common
14:21:56	FromGitter	<alehander42> not a big deal, I just define endl for my projects and interpolate it, I'll use the "" for " and use fmt extended literal always
14:22:15	Araq	"{a}\n{b}".fmt
14:22:33	Araq	can be your default syntax
14:24:05	Araq	alternatively we could make fmt interpret the \ differently but I dunno
14:24:09	FromGitter	<alehander42> hm, that's nice too
14:25:25	FromGitter	<RedBeard0531> I thought you had a plan for fmt/fmtr to handle that?
14:25:48	Araq	RedBeard0531: you had one, never got any PR from you
14:25:50	FromGitter	<alehander42> still I think some special handling for \ would be perfect , as one way to do something in 95% of cases (fmt".." instead of fmt function/method) will be best
14:26:26	Araq	we should really figure out this thing before releasing 0.18
14:26:39	Araq	also the lookup rules for the format protocol seem fishy
14:27:07	FromGitter	<RedBeard0531> Yeah, I found out that macros run too late for the difference from nnkStrLit vs nnkRawStrLit (or whatever the right names are) to matter
14:27:31	FromGitter	<RedBeard0531> I was hoping that would Just Work
14:27:56	Araq	well it's not much work to implement the lexing step that deals with \
14:28:10	Araq	but we need to figure out what to do.
14:28:43	Araq	the current way is super consistent but apparently non-intuitive
14:29:00	FromGitter	<RedBeard0531> Is there already a function that does nim-style escaping/unescaping?
14:29:09	Araq	(hey, dom96, see? consistency can produce *worse* results)
14:29:50	Araq	RedBeard0531: Nim is written in Nim, copy it from lexer.nim to be on the safe side
14:29:51	FromGitter	<RedBeard0531> It's not consistent with python's f-str vs fr-str :)
14:30:25	Araq	nor is the rest of Nim consistent with Python. strange argument.
14:30:46	dom96	Can't we just special case r""?
14:30:56	dom96	and get rid of this rule that blah"" is also a raw string literal?
14:31:00	FromGitter	<RedBeard0531> Sure, I just meant is there already one exposed in stdlib. I'll take your answer to mean no.
14:31:12	FromGitter	<GULPF> +1 dom96
14:31:28	Araq	dom96, and what about re"\d+" ?
14:31:35	Araq	or pegs?
14:31:49	Araq	there is a reason for why the things are as they are
14:31:50	dom96	Turn it into a macro and have it emit a raw string literal? :P
14:32:09	Araq	since fmt is already a macro, why not fix that instead?
14:32:11	dom96	yeah, nah, syntax highlighting that would be impossible
14:32:34	dom96	fixing fmt has the same problem
14:32:36	FromGitter	<RedBeard0531> Could the type system handle it if compile-time functions could take either raw or cooked strings?
14:32:54	FromGitter	<GULPF> fmt will need special syntax highlighting anyway though
14:33:03	FromGitter	<RedBeard0531> actually nevermind fmt needs raw strings
14:33:03	Araq	you can overload by ast kind but please let's not.
14:34:04	FromGitter	<RedBeard0531> fmt"{isSpace('\L')}" should work
14:34:27	FromGitter	<RedBeard0531> it is only in the "stringy" parts that escapes should be interpreted
14:34:58	Araq	I don't think so, char/string literals within fmt do smell.
14:35:32	FromGitter	<RedBeard0531> Sure its weird, but between the {} should be Plain Old Code
14:36:28	dom96	hm, that shows a limitation
14:36:37	FromGitter	<RedBeard0531> And sometimes its really useful: fmt"""blah {someSeq.join("\n")} blah"""
14:36:39	dom96	Technically everything inside {...} in a fmt"" should be treated as Nim code
14:36:56	dom96	which implies fmt"" should be implemented in the compiler :P
14:37:27	Araq	wrong, we can make fmt interpret \ outside the {} as we see fit
14:37:45	Araq	if we want to, no compiler magic required.
14:37:57	Araq	I don't think it's necessary though.
14:38:04	Araq	but be my guest.
14:38:11	dom96	well I dunno
14:38:16	dom96	I still haven't had a chance to test fmt""
14:38:21	dom96	because I was hoping we would rename it to f""
14:38:40	FromGitter	<RedBeard0531> IIRC In python f-strs you don't need to use """ to embed a ". But I don't see that working in nim without compiler support.
14:39:13	FromGitter	<RedBeard0531> To me that's not enough justification for compilerizing it. But nimsuggest support inside {} is :)
14:39:15	dom96	as far as this issue goes, either we embrace Nim or not
14:39:52	Araq	I will tell what does not embrace Nim. a shitty special lexer rule for "f strings"
14:40:19	dom96	Presumably, currently, char escapes in fmt"" are not intuitive because people are applying their knowledge of Python
14:40:44	FromGitter	<RedBeard0531> Not just python
14:40:51	euantor	I'd expect everything in an `fmt""` string to act the same as any other Nim string, unless isnde a `{}` context
14:40:51	Araq	no, it's that \n is more useful as a newline than as backslash-N
14:40:55	euantor	Same as in C~
14:40:59	euantor	*C#
14:41:17	FromGitter	<RedBeard0531> I think that is how every other language with string interpolation works
14:41:27	dom96	yeah, and without compiler support that's impossible, no?
14:42:13	FromGitter	<RedBeard0531> no, the macro can just unexcape before calling newStrLit
14:42:20	dom96	I can't write: fmt"{toUpper("\n")}"
14:42:28	dom96	or can I?
14:42:39	Araq	you can't because of the "
14:42:40	dom96	will it expand to toUpper("\n")?
14:42:43	FromGitter	<RedBeard0531> You need """ but then it would work
14:42:43	Araq	not because of the \
14:43:12	Araq	neither can you write String.format("{toUpper("\n")}") in C#
14:43:16	Araq	for the same reasons.
14:43:31	dom96	but I can in Python, no?
14:43:40	FromGitter	<alehander42> I agree with Araq it's about usefullness, not about Python users expectations
14:44:34	Araq	>>> f"{0}"""
14:44:34	Araq	'0'
14:44:38	dom96	another approach to subtly hinting people towards not using raw string literals is to rename this to `format` and make all examples: "{x}".format()
14:44:49	Araq	no idea what Python's doing here
14:45:00	FromGitter	<alehander42> and about consistency, if we need 3 special patterns (use fmt"..", but use fmt(..) for some escaped stuff or best, use var a = .., fmt"..{a}" for some joins because we can't join inside), that's not good
14:45:34	Araq	it is good to have a consistent language at the cost of a stdlib you have to learn.
14:45:41	FromGitter	<GULPF> handling `\` specially in fmt is brutally inconsistent though
14:45:54	FromGitter	<RedBeard0531> I think they made f"" and fr"" strings a new type of string in the lexer, at the same layer as r"" strings.
14:46:01	euantor	@Araq The modern C# way is `$"{toUpper("\n")}"`
14:46:07	euantor	And I believe that will work
14:46:25	dom96	Swift uses "blah: \(myVar)" as far as I can tell
14:46:54	Araq	GULPF: we can error out though and require {\} for literal backslashes
14:46:56	dom96	but indeed, Nim's philosophy is not to build these things into the language
14:47:41	Araq	Python's f-strings are broken
14:47:42	euantor	Yep, works as I thought it did. The C# compiler translates it to `string.Format("Hello {0}", "test".ToUpper())` https://www.irccloud.com/pastebin/4936Nbk9/
14:47:56	Araq	>>> f"{0}""908"
14:47:56	Araq	'0908'
14:48:08	dom96	https://www.python.org/dev/peps/pep-0498/#escape-sequences
14:48:08	Araq	where is my quote in there?
14:48:19	Araq	try it with Python 3.6.
14:48:28	dom96	That's just two string literals, side-by-side
14:48:42	Araq	oh, wtf
14:48:44	dom96	which Python concatenates
14:48:49	dom96	implicitly
14:48:53	Araq	I see, thanks
14:48:56	FromGitter	<alehander42> yep
14:49:16	euantor	@dom96 Yes, Swift uses `"\(var)"` which I personally don't like. Kotlin uses `"$var"` or `"${var}"` for more complex cases: https://kotlinlang.org/docs/reference/idioms.html#string-interpolation
14:49:19	Araq	yeah, I remember this misfeature of Python
14:49:36	Araq	euantor, it's about backslashes and quotes
14:49:44	Araq	not about dollar vs braces vs whatever
14:49:51	FromGitter	<alehander42> ok, why can't fmt implement special escape char handling itself, it's a macro, why would it need language level fixes
14:49:53	Araq	we settled for the curlies.
14:50:11	Araq	alehander42: yes, it can do that.
14:50:28	euantor	Curlies are for the best, was just wondering why dom posted about what Swift does
14:50:38	Araq	but I dunno whether it's a good idea
14:50:44	dom96	so, see in Python: fr"" is a raw f-string
14:50:51	dom96	whereas f"" is a f-string
14:51:08	Araq	I know.
14:51:12	dom96	that would be a weird rule for Nim
14:51:31	FromGitter	<RedBeard0531> But is seems like the most useful API
14:51:40	FromGitter	<GULPF> not for re
14:51:54	FromGitter	<RedBeard0531> Since 99% of the time you want your \n to mean newline new slash n
14:51:59	Araq	do we need "raw" f-strings though? surely I can use fmt""" """ instead
14:52:12	Araq	and make fmt distinguish between triple lit strings or not
14:52:29	Araq	the macro system allows for that
14:52:36	FromGitter	<RedBeard0531> You need """ to embed a " inside the {}
14:52:48	dom96	I don't think hacking fmt to treat fmt"" as a non-raw string literal sets a good precedent
14:53:10	FromGitter	<alehander42> it doesn't make sense for fmt to treat fmt"" as a raw string literal
14:53:19	Araq	dom96, for now it seems the most pragmatic solution though
14:53:27	Araq	and 'fmt' is all about getting shit done
14:53:33	Araq	and not about purity.
14:53:36	FromGitter	<alehander42> most of the time people format text or echo stuff with it
14:53:47	FromGitter	<RedBeard0531> That said, I don't know of any use cases for fmtr...
14:53:51	dom96	That's another rule for syntax highlighters to follow
14:54:06	FromGitter	<RedBeard0531> They should already special case fmt""
14:54:20	Araq	not really as far as the highlighter is concerned it can treat it like the Nim compiler does
14:54:31	Araq	fmt"" is an extended raw string literal.
14:54:43	dom96	hrm
14:55:18	FromGitter	<RedBeard0531> I'd rather a highligher be fmt-aware and highlight the stuff inside {} as code rather than a string
14:55:52	dom96	It won't be so easy
14:55:56	FromGitter	<RedBeard0531> That is what good highlighters do in python fstrs or js `${strs}`
14:56:02	dom96	The stuff inside {} won't exactly be the same as normal Nim code :\
14:56:18	FromGitter	<alehander42> exactly, highlighters for other languages with this kind of interpolation do work
14:56:22	FromGitter	<RedBeard0531> why not?
14:56:34	miran	Araq: re: "where is my quote in there? (f"{0}""908" > '0908') - this works as it should
14:56:56	dom96	well I suppose " is the only special case
14:57:08	miran	f"{0}" is first string, "908" is concatenated to it
14:57:25	Araq	miran, implicit string concats are dumb. ("a", "b" "c") # yay, typo lead to a bug
14:57:28	Yardanico	RedBeard0531: fmt supports some special stuff
14:57:37	Yardanico	it's not just plain string interpolation
14:57:45	miran	oh, now i see dom96 already answered that. sorry for duplication
14:57:57	Yardanico	it's possible to highlight it properly, but it requires some work
14:57:59	Araq	that should have fixed that bullshit for Python 3
14:58:18	Araq	only C requires it because it has no string concat operator that can work at compile-time
14:58:33	FromGitter	<RedBeard0531> "\"" is much more common (and logical) than """"
14:58:49	dom96	yeah, if " is the only special case then it matches Python behaviour as well
14:59:12	miran	that works nice when you break a long string in couple of lines of code
14:59:33	*	c0ntribut0r quit (Ping timeout: 265 seconds)
14:59:36	Araq	miran, a + also works nice.
14:59:44	*	c0ntribut0r joined #nim
14:59:45	miran	it is uglier
14:59:48	dom96	Araq: I do still get annoyed that I have to separate by & every time though :P
15:00:03	miran	just like \ line breaks are ugly
15:00:15	Yardanico	miran, I wouldn't agree that "a""b" is better than "a" + "b" :)
15:00:22	Araq	miran, I beg to differ, clarity cannot be ugly.
15:00:44	miran	Yardanico: i wouldn't agree that "a" + "b" is better than "ab" :P
15:00:54	dom96	okay, so since we're making fmt"" even more like python
15:01:01	dom96	Why not go all the way and change it's name to 'f'?
15:01:04	Araq	dom96, no, we are not.
15:01:12	Araq	we make it more useful for the common cases.
15:01:16	miran	dom96: why are you insisting on it so much? :D
15:01:39	dom96	because I love it
15:02:05	miran	as i said on github - i love it too in python. but in nim it feels wrong
15:02:15	dom96	Araq: And we do that by treating fmt"" as a non-raw-string literal
15:02:15	Yardanico	then even more people will use f without space
15:02:22	dom96	which puts us closer to Python semantics
15:03:03	*	kivyion quit (Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/)
15:03:11	Araq	we don't do anything like that. right now the proposal is to make fmt parse backslashes
15:03:11	*	yglukhov quit (Remote host closed the connection)
15:03:27	miran	can this (non)-raw discussion be supplemented with some examples?
15:03:36	Araq	much like regexes parse backslashes
15:03:37	dom96	Araq: What's the difference?
15:03:56	miran	what would `echo fmt "{4+3}\n{9+5}"` print?
15:03:57	dom96	The change in behaviour is the same
15:04:25	FromGitter	<RedBeard0531> mirin: https://github.com/RedBeard0531/icemonim/blob/master/icemonim.nim#L192 should work with fmt"" rather than requiring "".fmt
15:05:06	FromGitter	<RedBeard0531> 714
15:05:09	*	pwntus quit (Ping timeout: 264 seconds)
15:05:36	FromGitter	<RedBeard0531> shift-enter didn't work. pretend there is a real newline in there
15:05:38	*	yglukhov joined #nim
15:05:48	Yardanico	you don't need to use it like that
15:06:06	Yardanico	fmt "this should work too"
15:06:08	Yardanico	fmt("or this")
15:06:12	Yardanico	but I'm not sure :)
15:06:17	Araq	so ... yeah, there is precedent, regexes parse backslashes in their own way, fmt can parse backslashes too
15:06:39	Araq	mimicing Nim's backslash rules
15:06:41	*	yglukhov quit (Read error: Connection reset by peer)
15:06:50	dom96	Once again, why not rename to 'f'?
15:07:08	Araq	bluenote worked with 'f' and it did bite.
15:07:10	*	yglukhov joined #nim
15:07:17	dom96	how?
15:07:23	miran	2. what would `echo fmt"text {"more\ntext"} bla"1 print?
15:07:31	Araq	lots of reasons.
15:07:49	Yardanico	dom96, people will say "fuuu! nim copies everything from python! lame!!" :D
15:08:07	miran	dom96: 'f' is the most common one letter function name
15:08:10	Araq	proc generic[T] = var t = 9; echo f # typo not caught
15:08:39	dom96	Araq: Why is that not an error?
15:09:00	dom96	miran: Do you have data to back that claim? :P
15:09:15	Araq	proc transform(f: proc()) = echo f"{8}" # type error?
15:09:41	miran	dom96: you cannot go around telling people "stop using `i` as a variable in iteration, now it has a special meaning in nim"
15:09:42	Araq	this one works thanks to OR I think but I'm sure I can create a case where it hurts
15:10:07	Araq	dom96, 'f' is in scope, no error.
15:10:12	dom96	miran: That's not what I'm doing. That variable is indeed common, but 'f', I don't see it much.
15:10:45	dom96	Araq: 'f' takes arguments, how can it not be an error?
15:10:45	Araq	dom96, it will caught at generic instantiation time but it's meh
15:11:03	miran	dom96: then you're not looking at much mathematical/physics codes, nor functional programming
15:11:05	Araq	it is a "late" error.
15:11:29	*	yglukhov quit (Ping timeout: 248 seconds)
15:11:31	Araq	because in plenty of contexts Nim does symbol bindings without type disambiguations
15:11:46	Araq	as types are not yet available.
15:13:08	Araq	that said, I like 'fstr' better as 'fmt' as a name
15:13:27	Araq	or maybe 'fs'
15:13:34	FromGitter	<GULPF> both `fmt` and `fstr` are so long :(
15:13:40	*	c0ntribut0r quit (Ping timeout: 260 seconds)
15:13:44	Araq	'fs' then? :-)
15:13:56	dom96	Araq: Did you ever fix VS code's search?
15:13:59	subsetpark	fmt is pretty damn short, just to say
15:14:00	dom96	now it's failing for me...
15:14:13	Araq	what do you mean?
15:14:19	Yardanico	dom96, ah, when we talked about ripgrep?
15:14:26	Araq	yeah.
15:14:34	*	c0ntribut0r joined #nim
15:14:37	FromGitter	<GULPF> @subsetpark not compared to string interpolation in other languages
15:14:37	dom96	yeah
15:15:10	Yardanico	dom96, "search.useRipgrep": false
15:15:14	Yardanico	in your settings in vscode
15:15:18	FromGitter	<GULPF> 'fs' is better, but IMO it should be a 1-character name
15:15:33	Yardanico	all 1-character names are too common :)
15:15:42	Araq	can't take this seriously, sorry
15:15:52	subsetpark	Let's not overstate the case here. Python didn't even *have* f-strings up until about a week ago. Myself and the rest of the python-speaking world have been getting along just fine with "".format() for a long time.
15:15:53	miran	1-char name? big no from me
15:16:06	Araq	will name you GUL until you see the light :P
15:16:15	Yardanico	subsetpark, week ? wtf
15:16:16	dom96	btw
15:16:19	Yardanico	:P
15:16:20	FromGitter	<GULPF> haha
15:16:23	dom96	var fmt = 42; fmt"..." works fine
15:16:29	miran	Araq: GUL is too long, why not G?
15:16:30	dom96	as does defining a `proc fmt`
15:16:44	miran	G""
15:17:00	dom96	so I'm still unconvinced that this will be a problem
15:17:07	FromGitter	<GULPF> you have no right to complain, the compiler is full of 1-character names after all ; )
15:17:11	dom96	You can still use 'f' as an identifier to your hearts content
15:17:28	Yardanico	◍"this is better" :D
15:17:33	Araq	GULPF: not because I'm too lazy to type.
15:17:38	Yardanico	(and it's actually possible lol)
15:17:41	Araq	because I'm too lazy to invent stupid names.
15:18:03	Araq	actually I'm also to lazy to read the result, so ok.
15:18:13	Araq	point goes to you.
15:18:40	dom96	Araq: Also, your first example doesn't compile
15:18:56	Araq	dom96, it does if you have one 'f' in your scope
15:19:11	Araq	which is the context we're talking about.
15:19:37	FromGitter	<GULPF> what about using an operator like `@`? C# uses `$` so it's not completely alien
15:19:54	dom96	oh right, well, that already happens a lot with generics
15:20:00	Araq	operators are reserved for common operations.
15:20:04	dom96	IMO it's a non-issue
15:20:25	Araq	and "common" is defined as "what Araq considers important"
15:20:37	dom96	I could import some package that exports 'f' too and have the same problem
15:20:43	dom96	and I wouldn't even know
15:20:44	Araq	since Araq doesn't even like format strings an operator is off the table :P
15:21:01	Yardanico	argh, again unicorn on github
15:21:03	FromGitter	<GULPF> well then :P
15:21:10	Yardanico	https://twitter.com/githubstatus/status/951473930031779841
15:21:15	dom96	yeah, gist is unicorning
15:21:19	Yardanico	not only gist
15:21:23	Yardanico	github is unicorning for me
15:22:43	*	miran left #nim (#nim)
15:22:53	dom96	Pastebin it is I guess
15:23:02	dom96	https://pastebin.com/BnD8dxqD
15:23:09	dom96	No issues there
15:23:17	FromGitter	<RedBeard0531> That is worded strangely. Like they are waiting for people to report to them that their site is down rather than having monitoring in place.
15:23:38	dom96	and this works too https://pastebin.com/7ae2xs4f
15:23:42	Yardanico	yay they updated their status https://status.github.com/messages
15:23:47	Yardanico	" We are investigating reports of service unavailability."
15:24:00	subsetpark	the web is poison, people!
15:24:13	dom96	wow, so much red in the past
15:24:23	Yardanico	dom96, it wasn't there before they changed current status
15:24:29	Araq	ok, here is the solution
15:24:35	Araq	an unary % operator!
15:24:42	dom96	Another reminder that we should come up with a way to store issues/PRs/etc in the git repo as well
15:24:48	Araq	% is already the format operator
15:24:51	Yardanico	Araq, hmm
15:24:59	Yardanico	%"{myvar}" not that bad
15:25:02	Araq	and %"{0}\n" does the right thing
15:25:07	Yardanico	well I mean it's good :)
15:25:11	Araq	since it's not a raw string literal
15:25:12	subsetpark	That makes a lot of sense...
15:25:35	Yardanico	Araq, will %"{a[\"b\"]}" work? yay
15:25:45	FromGitter	<GULPF> +1 from me
15:26:24	dom96	hah
15:26:34	dom96	that's genius
15:26:36	FromGitter	<RedBeard0531> It would be weird if the escaping applied inside the {}
15:26:53	dom96	and we're back to square one :P
15:27:00	FromGitter	<RedBeard0531> but still seems better than status quo...hmm
15:27:18	Yardanico	:D https://strawpoll.com/hy366wp8
15:27:34	Yardanico	sorry, no "other" option today
15:27:36	dom96	ooh, strawpoll got a redesign
15:27:46	dom96	Now I'm reminded of our livestream polls
15:27:47	Yardanico	that's another strawpoll afaik
15:27:55	dom96	Hrm, maybe I'll do a livestream this weekend
15:27:58	Yardanico	http://www.strawpoll.me/ this is the one we've used in the past
15:28:08	dom96	oh, interesting
15:29:52	dom96	I didn't realise that operators in front of string literals don't create a raw string
15:32:27	Araq	yeah, let's go with % I doubt it can be topped
15:32:43	dom96	Challenge acce-- nah, I doubt it too
15:34:24	dom96	I wonder how nitely's regex implementation is going (that was nitely wasn't it?)
15:39:08	Yardanico	yay, github works
15:41:16	Yardanico	argh, my eyes are bleeding https://github.com/bung87/gpx2baidumap-array/blob/master/main.nim
15:42:26	dom96	this is why we need nimpretty
15:42:33	*	azur_kind joined #nim
15:44:43	Yardanico	actually I did a few random PRs to small nim projects to make them look prettier :)
15:45:26	livcd	did i see nimpretty here ? :D
15:46:16	Yardanico	dom96, I bet changelog (including closed issues) will be quite big for 0.18.0 :P
15:47:24	Yardanico	it's very satisfying to read big changelogs btw (unless you have to fix a ton of deprecation warnings)
15:50:17	Yardanico	also - semaphoreci is faster even on small projects (because of nim compiler compilation), it takes only ~1:30 to build nim and nimble: https://semaphoreci.com/yardanico/nickel/branches/master/builds/3
15:50:48	Yardanico	(best test suite ever - check if program compiles)
15:54:31	*	BitPuffin quit (Remote host closed the connection)
16:09:36	FromGitter	<Varriount> semaphorici?
16:11:13	*	der joined #nim
16:12:40	Yardanico	SemaphoreCI :)
16:15:03	*	der quit (Remote host closed the connection)
16:15:17	*	derlafff quit (Remote host closed the connection)
16:15:36	*	derlafff joined #nim
16:15:36	FromGitter	<alehander42> unary `%` is smart :D
16:15:48	*	endragor quit (Remote host closed the connection)
16:15:52	*	miran joined #nim
16:19:35	*	PMunch quit (Ping timeout: 260 seconds)
16:24:07	Araq	even better, %r"\n" gives a raw string
16:24:16	Araq	no special casing in the macro required
16:24:26	Araq	everything works out beautifully
16:24:58	Yardanico	Araq, will you push it today? :)
16:26:27	*	arnetheduck quit (Ping timeout: 240 seconds)
16:31:14	*	floppydh quit (Quit: WeeChat 2.0.1)
16:31:26	Araq	yes
16:33:22	*	endragor joined #nim
16:35:23	*	zahary joined #nim
16:37:41	*	endragor quit (Ping timeout: 256 seconds)
16:43:27	*	yaiyan is now known as Yaiyan
16:47:00	miran	if somebody could create a forum post to showcase the new usage of %, that would be great
16:47:38	miran	on the other hand, those kinds of things should be somewhere in documentation/manual, right?
16:47:49	Araq	lol why? %"{foo}" is the syntax
16:48:09	Araq	the manual contains the lexing rules, yes
16:49:15	Araq	I can replace 'fmt' with '%' in strformat, that's the whole patch
16:49:27	miran	how does it behave with those \n inside, and how outside of {}?
16:49:40	*	azur_kind quit (Remote host closed the connection)
16:50:00	miran	can i do %"{3.333333:.2f}"?
16:50:45	miran	can i do %"{2.5:>10}"?
16:51:33	Araq	why not?
16:51:51	Araq	identifier"string" is an extended raw string literal
16:51:55	Araq	%"string" is not
16:51:59	miran	i don't know why not. i don't know what works and what doesn't
16:52:02	Araq	so \n is parsed differently
16:52:25	Araq	according to Nim's lexing rules
16:52:34	Araq	everything works out
16:53:40	*	azur_kind joined #nim
16:58:39	*	azur_kind quit (Remote host closed the connection)
17:03:06	c0ntribut0r	Hello guys! Let me distract you from that "%" thing :) Is there any way in nim to tell compiler to use tail call optimization?
17:08:06	Araq	c0ntribut0r, no, you can enable it for GCC though
17:08:21	Araq	independently of the other optimizer settings afaict
17:09:15	Yardanico	there's also tailcallelim pass in clang https://llvm.org/docs/Passes.html#tailcallelim-tail-call-elimination
17:09:31	Araq	now ... what do we do with the "format" protocol?
17:09:40	Yardanico	Araq, what is wrong with it?
17:10:11	Araq	it simply assumes that format(T, string) yields void and means "append" semantics
17:10:17	c0ntribut0r	Thanks everyone
17:10:26	Araq	addFormat would be clearer
17:12:12	dom96	Araq: What's wrong with that?
17:19:42	*	darithorn joined #nim
17:21:11	*	claudiuinberlin quit (Quit: Textual IRC Client: www.textualapp.com)
17:24:51	FromGitter	<krux02> Araq: can you give me a few seconds of your time?
17:25:05	FromGitter	<krux02> I try to understand the compiler internal type ast
17:25:32	Yardanico	OH MY EYES https://github.com/nim-lang/Nim/blob/425f221440cbd453591d0fca9eaf93671d5f6d62/lib/pure/strformat.nim
17:25:35	FromGitter	<krux02> here is the AST I am currently working with http://ix.io/E1Z
17:27:09	FromGitter	<krux02> or here with line numbers: http://ix.io/E1Z/Nim
17:28:31	FromGitter	<krux02> I try to extract Vec[4,float32] as a node, somehow.
17:29:25	FromGitter	<krux02> the only position in this AST where I can see a 4 followed by a float32 is in line 75
17:32:03	dom96	Araq: Why was readChar deprecated?
17:32:26	*	sendell quit (Remote host closed the connection)
17:33:39	dom96	It was deprecated here and I don't get it: https://github.com/nim-lang/Nim/commit/794d36cf31d8f7debc2ede1f7792fdbcc0a1ef5e
17:36:04	Araq	krux02: yeah it's messy, usually I only look at the "implementation" part
17:36:19	dom96	"readChar shouldn't be used for anything as it's too slow (usually needs to acquire a lock)."
17:36:26	Araq	(the object arr: array[] thing you refuse to base your analysis on)
17:36:29	dom96	Despite this I would argue it's still useful
17:36:55	Araq	dom96, it also uses the IOError exception as the EOF marker
17:37:02	Araq	which is also not nice IMO.
17:37:14	dom96	The reason I ask is because of this https://stackoverflow.com/questions/48186397/nim-readchar-deprecated-what-to-use-instead
17:37:36	Araq	if readBuffer(addr myChar, 1) == 1
17:37:41	Araq	or something like that
17:38:27	dom96	look at the code
17:38:36	dom96	replacing that function with this is bleh
17:38:40	dom96	and it's unsafe code (!)
17:41:26	Yardanico	Araq, ohhh, % has the same signature as %from json, so % from json is being implicitly picked up by default
17:42:07	Yardanico	hmm, I don't get it
17:42:55	Yardanico	it works fine in separated code
17:42:58	Yardanico	but not in my project :)
17:43:31	Yardanico	oh, here it is
17:44:21	Yardanico	Araq, seems like I shouldn't use unindent this way :) https://gist.github.com/Yardanico/f6f9f5942f936d1db9cfd33f3257f330
17:44:39	Yardanico	and this code compiles
17:44:53	Yardanico	and echoes "{a}\n{b}\n" :)
17:46:58	Araq	Yardanico: argh...
17:47:32	Araq	any obvious solutions?
17:48:02	Araq	omg that was stupid
17:48:13	dom96	omg lol
17:48:34	Yardanico	so as I understand unindent is called first, and only after json.% is called
17:49:18	*	Pisuke joined #nim
17:49:39	*	MyMind quit (Ping timeout: 256 seconds)
17:49:52	FromGitter	<Yardanico> *after that
17:50:23	*	azur_kind joined #nim
17:51:05	FromGitter	<krux02> Araq: what does the last child of a type semantically mean?
17:51:12	Araq	Yardanico: +obj.field # how do think this is parsed?
17:51:25	Araq	krux02: the instantiation
17:51:38	Araq	Yardanico: %x.f is the same
17:53:26	FromGitter	<krux02> Araq: getTypeInst recrusevely taks the last child of the last child until termination
17:53:40	FromGitter	<krux02> that is causing the problem I have
17:54:16	Araq	quite possibly
17:54:22	Araq	try it without the loop?
17:54:38	Araq	hello ...? I need a solution asap
17:54:50	Araq	strformat.% bites json.%
17:55:19	FromGitter	<krux02> well the loop was introduced intentionally
17:55:25	FromGitter	<krux02> with a patch recently
17:55:51	FromGitter	<krux02> I check when it happened
17:56:02	dom96	Araq: I have no ideas
17:56:08	dom96	There isn't really a solution
17:56:16	dom96	Either we use a different operator
17:56:25	dom96	or ... there is no other solution
17:56:26	Araq	which one?
17:56:32	Araq	which operator to pick instead?
17:56:44	dom96	\ ?
17:57:00	Araq	unary '&' ? because it translates into a concat expression?
17:57:24	Araq	&"{foo}\n" # still looks nice
17:57:38	dom96	I don't like that: "Hello" & &"{foo}"
17:57:58	Araq	hmm but it hardly comes up
17:58:15	Araq	since you would embed the concat into the format string, that's its point after all.
17:58:15	dom96	it will come up if I need to wrap the strings
17:58:35	Araq	I said "hardly", not never.
17:59:28	miran	back to 'fmt'? :D
17:59:49	Araq	*& but then it's 2 chars and somebody's fingers will start to bleed
18:00:01	dom96	two shifted chars as well...
18:00:05	Yardanico	miran, nah, it'll still be a operator I think
18:00:19	Araq	the operator solution is too elegant
18:00:21	Yardanico	people like me will need to carefully read manual :P
18:00:30	dom96	we could also rename json.%
18:00:33	miran	how come we don't have bitwise & and |?
18:00:39	Yardanico	miran, we do
18:00:40	dom96	well, we actually can't
18:00:40	Araq	we need to keep it but pick an operator that is not yet used.
18:00:42	dom96	It's canon
18:00:42	Yardanico	and and or
18:00:46	miran	i guess it was readability
18:00:51	Araq	dom96: was about to say that :P
18:01:22	miran	Yardanico: i'm talking about exactly & and | as keywords, not `and` and `or`
18:01:34	Yardanico	keywords? :D
18:01:46	miran	and now suddenly `&` is ok for something more common
18:02:07	miran	and it is something which is not as widespread as bitwise &
18:02:38	Araq	I like *& fwiw, it produces zero or more concats
18:02:53	Araq	*&"{foo}\n"
18:03:02	*	yglukhov joined #nim
18:03:03	Araq	meh. I dunno
18:03:08	miran	i don't see connection between `*&` and string formatting
18:03:08	dom96	-1 from me
18:03:18	dom96	\"{foo}\n"
18:03:22	dom96	Inspired by Swift
18:03:31	dom96	not sure if that's even possible
18:03:35	Araq	dom96: I find that harder to read
18:03:43	Araq	it is possible.
18:03:52	miran	?"{bla}"
18:04:02	Yardanico	!@#$
18:04:05	Araq	"lambda foo -- what?"
18:04:17	dom96	yeah, reminds me of lambda too
18:04:38	miran	to be honest, ? makes more sense to me than &
18:04:39	dom96	it does contain code though, so it sorta makes sense
18:04:52	miran	you're "asking" to evaluate string
18:05:20	dom96	@"{bla}" maybe?
18:05:34	dom96	although with seq using it, it's weird
18:07:33	*	yglukhov quit (Ping timeout: 264 seconds)
18:10:22	*	S1tiSchu quit (Read error: Connection reset by peer)
18:11:07	miran	it should be a symbol not used elsewhere, so that leaves out: @#$%^&*()-+=/<>\
18:11:27	miran	remaining: ` ~ ?
18:11:54	dom96	what about ! ?
18:12:03	Yardanico	macros module?
18:12:07	Yardanico	but ! is deprecated anyway :)
18:12:12	Yardanico	(in macros)
18:12:12	miran	` is too similar to ', ~ is too similar to - (remember those don't use `l` as a name warnings?)
18:12:17	dom96	yeah, its use in macros isn't very useful
18:12:51	miran	`!` seems logical
18:13:22	dom96	I think we've kind of been loosely keeping `!` free for others to use
18:13:39	*	devdri joined #nim
18:13:57	Yardanico	and people still say that nim is unstable, look at this: https://gist.github.com/Yardanico/adb9315d3b77df3998fa195453126cf7 I have a DSL in my project which also uses async, so C code is compilcated as hell, but it works :)
18:14:26	*	devdri quit (Client Quit)
18:15:22	Yardanico	(it's compiled in release mode here)
18:15:26	*	yglukhov joined #nim
18:16:05	miran	ok, if ! should be used for other stuff, what do you guys think about `?` ?
18:16:29	*	yglukhov quit (Read error: Connection reset by peer)
18:16:53	miran	?"{3+4}"
18:17:01	*	yglukhov joined #nim
18:17:14	FromGitter	<RedBeard0531> I still like fmt. It's also easier to do the right thing internally if it is a name rather than an operator
18:17:58	miran	`fmt` is fine by me, but suddenly there was "it has to be an operator":)
18:20:41	*	Trustable joined #nim
18:25:03	FromGitter	<RedBeard0531> It also seems like it would be generally useful to expose equivalent logic from getString in macros.nim. Once that is done, it becomes easy to make fmt do the right thing.
18:27:54	*	Sembei joined #nim
18:28:02	*	Pisuke quit (Ping timeout: 255 seconds)
18:28:47	dom96	Operator makes sense because it won't require parsing of the backslashes in the string
18:29:12	*	Vladar quit (Remote host closed the connection)
18:29:22	FromGitter	<RedBeard0531> But that makes it impossible to avoid parsing backslashes inside the {}
18:30:00	*	azur_kind quit (Read error: Connection reset by peer)
18:30:21	FromGitter	<RedBeard0531> And I think unescapeStr is a generally useful function to have exposed, which takes the hard part out of fmt itself
18:30:38	*	azur_kind joined #nim
18:35:51	FromGitter	<RedBeard0531> strutils.unescape is almost right, but it only handles a subset of the \ escape codes. Can that be expanded to cover the rest and error on unknown escapes?
18:36:06	*	Jesin joined #nim
18:39:33	*	Vladar joined #nim
18:51:35	FromGitter	<GULPF> making it an operator means that doesn
18:52:02	FromGitter	<GULPF> *means that it doesn't need special handling of '\' though
18:52:40	FromGitter	<GULPF> it's much more elegant that any alternative
18:54:35	*	c0ntribut0r quit (Ping timeout: 240 seconds)
18:54:41	FromGitter	<GULPF> the problem is not that it's hard to implement with a normal identifier, the problem is that it breaks the consistency of the raw string syntax
18:55:26	*	yglukhov quit (Read error: Connection reset by peer)
18:55:36	*	azur_kind quit (Remote host closed the connection)
18:56:02	*	yglukhov joined #nim
19:04:02	*	devdri joined #nim
19:06:44	FromGitter	<RedBeard0531> But you don't actually want fmt to take a fully cooked string. You only want to parse escapes outside of the {}. You can't do that with an operator.
19:17:17	FromGitter	<GULPF> Hmm that means that operators named `\` can't be used inside the `{}`. That's pretty bad
19:21:55	*	Vladar quit (Quit: Leaving)
19:24:31	yglukhov	dom96: you around?
19:24:46	dom96	yes
19:27:28	FromGitter	<RedBeard0531> Gulph that is why you don't want fmt to be an operator.
19:29:14	Araq	RedBeard0531: use ?r"\n" to get the non interpreted backslashes
19:29:18	Araq	works.
19:29:28	Araq	(assuming we go for '?')
19:29:51	Araq	echo ?"{foo}"
19:30:00	Araq	I still prefer unary &
19:30:10	Araq	it's sugar for a concatenation
19:30:17	Araq	why not use & ?
19:30:43	miran	"abc" & &"{123}"
19:30:56	Araq	so what, nobody has proven that comes up
19:31:19	Araq	besides, I use that in Karax and it quickly becomes acceptable
19:31:38	FromGitter	<RedBeard0531> The key is that I think the most common use case wants *some* slashes escaped
19:31:48	miran	i don't see the connection between the symbol & and string formatting
19:31:57	Araq	&"abc{123}"
19:31:59	Araq	-->
19:32:13	Araq	"abc" & $123
19:32:29	Araq	how can you not see the connection?
19:33:00	miran	btw, what do you think about slightly changing how `echo` works? echo a, b, c now produces abc
19:33:15	miran	what about `,` produces one space?
19:33:40	Araq	don't see the point
19:33:55	miran	better connection is $"abc{123}"
19:34:00	Araq	for debugging it's usally echo a, " b: ", b
19:34:10	miran	let a = 10; b = 30
19:34:17	miran	echo a, b --> 1030
19:34:21	Araq	with string literals in between so you can have your spaces there
19:34:30	Araq	echo a, " ", b
19:34:42	Araq	echo a, ": ", b
19:34:53	Araq	echo "(", a, ", ", b, ")"
19:35:13	miran	you don't find it ugly/unnecessary?
19:35:34	miran	if i want concatenated, i would use & or sth
19:35:41	miran	usually i want them separated
19:36:06	Araq	yeah but I never want them separated by a space.
19:36:25	Araq	and so there is no good default.
19:36:25	FromGitter	<RedBeard0531> Any chance future.dump could be varargs? That seems more useful for a lot of cases
19:36:37	miran	echo "(", a, ", ", b, ")" -> this will be much nicer with fmt/f/%/&/? :D
19:36:43	Araq	also concentations are more expensive than a varargs mechanism
19:37:29	Araq	how do I "not" use the space in Python and avoid the allocations? ok, it's Python, nobody cares about allocations.
19:37:31	miran	echo ?"{a}{b}"
19:37:39	Araq	but Nim is systems programming.
19:37:54	miran	nice and simple :)
19:38:00	FromGitter	<RedBeard0531> Also, if command syntax could use named args, you could probably do echo a, b, sep=" "
19:38:07	Araq	hardly. it's disgusting
19:38:14	alexday	hi Araq , I just wanted to keep in touch with you. :P
19:39:07	FromGitter	<RedBeard0531> Is there a technical reason why command syntax can't support named args?
19:39:28	Araq	I think it would take the syntax past its breaking point
19:39:31	Araq	but I'm not sure.
19:40:14	FromGitter	<RedBeard0531> Breaking compilers is my specialty ☺
19:40:17	Araq	so ? or &
19:40:29	miran	ok, so no space between things in echo? :( i guess i'll have to use string formatting for that.... (as i find `a, " ", b` ugly)
19:40:29	Araq	can we decide on that? I want to work on something important.
19:40:40	miran	imho ?
19:41:00	miran	i stated my reasons above, but i can repeat them if necessary
19:41:15	Araq	miran, it's logical and with a bit of luck makes you use an even better separator
19:41:24	miran	what about `!`?
19:41:32	Araq	so that others can interpret the output of your programs
19:42:17	*	Yardanico_ joined #nim
19:43:18	Araq	miran, I think '!' should be some dangerous operation
19:43:39	miran	haha, ok
19:44:15	miran	then `?` as a shorthand for "can you evalute this string please?" :)
19:44:27	Araq	echo &"{foo} {bar}"
19:44:30	*	MyMind joined #nim
19:44:34	Araq	echo ?"{foo} {bar}"
19:44:47	miran	what is foo? what is bar?
19:44:55	miran	(translated)
19:45:04	*	Yardanico quit (Ping timeout: 255 seconds)
19:45:08	Araq	why does it matter? it uses your precious space
19:45:26	Araq	as the separator, that should be enough to make you happy :P
19:45:28	miran	&"{foo} --> "concatenate one thing?"
19:45:42	Araq	yes, nothing wrong with that
19:45:57	*	Sembei quit (Ping timeout: 240 seconds)
19:46:07	*	kier quit (Ping timeout: 248 seconds)
19:46:12	Yardanico_	4400 stars :)
19:46:19	miran	i just wanted to say that ? makes more sense to me
19:46:35	Araq	ok, now we only need some feedback from others
19:46:42	Araq	Yardanico_, !
19:46:51	Araq	what do you say?
19:46:53	FromGitter	<RedBeard0531> ! Would be a good operator to let you store openarray in a data structure, not just an arg☺
19:47:06	miran	well, dom96 is for `f` and nothing else :D :D
19:47:30	FromGitter	<data-man> echo <{foo} {bar}>
19:47:40	Yardanico_	Araq, honestly I don't encounter a lot of operators like ! or @ , so I'm fine with !
19:47:56	Araq	but ! is not in your list of options :P
19:48:00	Araq	it's ? or &
19:48:00	*	Yardanico_ is now known as Yardanico
19:48:27	FromGitter	<GULPF> ! > ? > & imo
19:48:34	Yardanico	Araq, well ? is a bit easier for me to type because it's closer to my fingers :P
19:49:19	FromGitter	<RedBeard0531> https://nim-lang.org/docs/pegs.html#%3F,Peg
19:49:58	FromGitter	<data-man> echo <{format}{me}"please">
19:50:09	miran	GULPF: agreed
19:50:24	Araq	data-man: we don't have angle brackets
19:50:28	*	kier joined #nim
19:51:28	Yardanico	data-man: nim tries to implement as much of stdlib without compiler changes as possible
19:51:34	Yardanico	that's why async is not built-in for example
19:51:39	Yardanico	so core language is not that complicated
19:52:36	Araq	echo +"{foo}"
19:52:54	Araq	unary plus anyone?
19:53:28	FromGitter	<data-man> Ё"Hi{Yo!}"
19:53:49	Araq	data-man: it has to be an operator
19:54:17	miran	Araq: maybe you can leave !! for something exclamatory, and have ! for this? :)
19:54:23	Araq	!\?&"{0}" # best compromise so far
19:54:46	miran	:D
19:54:47	*	Ven`` joined #nim
19:54:48	FromGitter	<RedBeard0531> Araq: would a PR to make fmt only unescape the string parts be rejected outright? I still think that is the right design, and I'm willing to write it.
19:54:51	Araq	the "bang backlash question mark amp" operator
19:54:59	Yardanico	data-man: cyrillic letters huh? :P
19:55:04	FromGitter	<RedBeard0531> aka the "fuck this shit" operator
19:55:13	*	gokr joined #nim
19:55:14	Yardanico	Araq, !@#$ is better maybe? :)
19:55:21	FromGitter	<RedBeard0531> capital s-sharp?
19:55:24	Yardanico	ah
19:55:27	Yardanico	# can't be used
19:55:51	Araq	RedBeard0531: I don't like the name 'fmt' tbh
19:56:03	FromGitter	<RedBeard0531> emoji happy face since it makes you happy when you use it?
19:56:16	FromGitter	<RedBeard0531> Sorry, I've got my troll hat on and it is hard to remove...
19:56:18	*	gokr left #nim (#nim)
19:57:00	Yardanico	let's bring some names from emojicode programming language
19:57:05	Yardanico	🐖
19:57:11	FromGitter	<RedBeard0531> Hmm I like fmt since it seems imediately obvious what it is for. but `fstr` also seems fine
19:57:18	Araq	can't we remove the string{lit} and then json.% for strings would be ambiguous with strformat.%
19:57:45	FromGitter	<RedBeard0531> although I think that only makes sense by reference to a python feature, which is kinda silly
19:57:48	FromGitter	<data-man> > "{fmt}"
19:57:58	Araq	so only the people who use both all the time need to do something like import json except %
19:58:28	Araq	data-man: unary > is rewritten to < it would work but it also would be weird
19:58:35	Yardanico	Araq, I think strformat will be pretty commonly used with json
19:58:50	Araq	ok
19:58:51	FromGitter	<data-man> > "{fmt}"
19:59:00	Yardanico	I mean because people will for example call json apis
19:59:08	FromGitter	<data-man> Two >
19:59:30	FromGitter	<RedBeard0531> interp""? a bit long I guess
19:59:35	Yardanico	well I mean I'm ok with any choice
20:00:15	FromGitter	<RedBeard0531> Oh, is `F` a common name? How about F""?
20:00:31	FromGitter	<RedBeard0531> it won't collide with `f` due to partial case sensitivity
20:00:48	Araq	yeah I know but it's not in NEP-1
20:01:00	Araq	and not that good to look at either IMO
20:01:24	Araq	also it means we need to use the complex implementation that cares about backslashes
20:02:09	FromGitter	<data-man> _{fmt} :-)
20:02:14	FromGitter	<RedBeard0531> I don't think so. I think it is just a matter of replacing the call to newStrLit(x) with newStrLit(unescape(x))
20:03:56	Araq	an operator feel more idiomatic to me. :P
20:05:18	FromGitter	<RedBeard0531> Yeah, but it does the wrong thing inside of {}. Also it would behave differently with """ strings
20:07:09	*	yglukhov quit (Read error: Connection reset by peer)
20:07:37	FromGitter	<RedBeard0531> I think python got the f- vs fr-string distinction right. The stuff between {} never interprets \ and f vs fr decides how to treat \ outside the {}. I'm hopeful we can make it work that well in nim.
20:07:56	*	yglukhov joined #nim
20:10:36	Araq	I think I'll go with unary +, it has no meaning for strings and easy to type
20:10:50	Araq	and that's the main selling point of 'fmt' in the first place
20:11:16	alexday	when I do a nimble install nimsuggest. why does it start cloning the Nim repo?
20:11:27	Yardanico	alexday, firstly - why do you install it like that?
20:11:30	Yardanico	secondly - it uses compiler
20:11:58	alexday	then how do you install?
20:12:17	alexday	with koch?
20:13:13	FromGitter	<data-man> ~"{a}{b}"
20:13:29	Yardanico	alexday, yes
20:13:37	alexday	ok
20:13:40	Yardanico	./koch tools will clone&compile nimble and nimsuggest
20:14:40	alexday	oh, Araq is the founder of nim :P . maybe he will mentor me someday :P
20:14:47	alexday	on writing languages
20:15:26	FromGitter	<RedBeard0531> ~ is the char lisp uses for formatting. but that always felt insane. D uses ~ for concat, so I guess it is kinda like &
20:16:41	FromGitter	<nitely> what about `%%"{abc}"`?
20:16:56	*	devdri quit ()
20:20:37	FromGitter	<data-man> @@"{a}{b}" ⏎ ->"{a}{b}" ⏎ !!"{a}{b}"
20:20:45	*	claudiuinberlin joined #nim
20:22:00	FromGitter	<RedBeard0531> would implementing `macro \`{}\` (fmt: string): untyped` work? The use of {} feels right with the syntax.
20:22:54	FromGitter	<RedBeard0531> Well I fail at markdown. I meant for it to be named {} with the striped syntax.
20:27:09	dom96	to be honest I dislike all operators except % :(
20:30:11	FromGitter	<data-man> sf"{a}{b}"
20:30:50	dom96	nitely: how's your regex package coming along?
20:31:02	dom96	also, %% might be a good idea
20:33:58	Araq	%%"{x}" this gets ugly fast
20:34:15	Araq	+"{x}" is growing on me
20:34:39	dom96	I think I just prefer fmt"" now
20:34:43	FromGitter	<nitely> dom96: almost there. I'll open source it this weekend
20:35:02	dom96	nitely: why not just pop it on to github now?
20:35:10	Araq	dom96, that doesn't work with \n though and would have to be special cased
20:35:20	Araq	and then it won't work with " inside
20:35:24	dom96	we were going to do that anyway
20:35:36	dom96	it shouldn't work with " inside
20:35:43	FromGitter	<nitely> dom96: you wanna try it or just check the code?
20:36:02	Araq	dom96, problem is wouldn't work with \" either
20:36:14	Araq	which is hard to argue for
20:36:18	dom96	nitely: I'm just afraid you'll disappear someday and we won't ever get to see your masterpiece
20:36:35	dom96	but I would like to try it
20:37:20	FromGitter	<nitely> dom96: ahaha I'm sure if I get hit by a bus someone in my family will release it XD
20:37:30	FromGitter	<data-man> +1 to try it :)
20:37:46	*	devdri joined #nim
20:37:57	dom96	still, you should release it right now, there is no reason not to :)
20:38:15	FromGitter	<nitely> problem is, it requires my package nitely/nim-unicodedb and there were a couple of things missing theere
20:38:35	*	yglukhov quit (Read error: Connection reset by peer)
20:39:18	*	yglukhov joined #nim
20:39:56	FromGitter	<nitely> mainly for matching unicode with shorthands: `\w`, `\d`, etc
20:40:19	dom96	why is that a problem?
20:40:28	dom96	it doesn't have to be working 100% to be released :)
20:40:46	FromGitter	<nitely> well, it's almost done now anyway
20:41:04	Yardanico	wow, it supports unicode?
20:41:05	FromGitter	<nitely> I'll try and release it in a couple hours
20:41:11	dom96	thanks :)
20:41:22	dom96	Yardanico: Kind of a must-have for regex engines nowadays
20:41:26	FromGitter	<nitely> my goal is to have unicode level 2 support (i.e: graphemes)
20:41:40	FromGitter	<nitely> but that will take much longer
20:41:59	Araq	release early, release often. and start with version 1.
20:42:11	Araq	bufixes produce version 1.1, 1.2 etc
20:42:22	Araq	graphemes produce 2.0
20:42:31	Araq	see Nim for how to not do things
20:42:48	Yardanico	yeah, if nim would be 17.2 no one would be complaining about 1.0 :D
20:42:51	Araq	version 1 should be "works for me"
20:42:56	enthus1ast	and if you see v1.1 wait for v1.2 ;)
20:43:43	*	Yardanico quit (Remote host closed the connection)
20:44:14	Araq	or go for a year-month release.
20:45:13	Araq	open source software is not a master thesis, there is no deadline
20:45:39	Araq	and the goals shift within time.
20:46:11	Araq	when you were born you couldn't even eat properly, your mother released you anyway
20:46:36	enthus1ast	but she had a deadline!
20:46:44	Araq	lol
20:47:11	Araq	ok, release any crap you have after 9 months
20:49:43	FromGitter	<data-man> A little bit about music https://github.com/lieff/minimp3
20:49:49	enthus1ast	or sometimes deadlines help to get the best out
20:52:15	miran	i've been afk for cca 1 hour. has any decision been made? :)
20:52:40	Araq	always remember beneath a steel sky. https://youtu.be/d-x0qghlvfE?t=569
20:54:27	Araq	miran, I went with an unary plus
20:55:22	miran	+"{it's better than &}"
20:55:44	Araq	the + looks a bit like an f :P
20:56:25	dom96	:/
20:56:28	FromGitter	<RedBeard0531> are you drunk or just have worse vision than me?
20:56:35	*	dom96 dislikes it
20:57:02	miran	:D
20:57:04	dom96	let's go back to fmt""
20:57:14	dom96	I guess that's what I get about complaining about that name...
20:57:26	miran	memo for the future :)
20:57:45	dom96	hah, nah, nothing will ever stop me fighting for what I believe in!
20:58:31	Araq	+ t f
20:58:32	FromGitter	<RedBeard0531> My offer to make fmt"" work correctly still stands :)
20:58:46	Araq	and f is a plus with some bow
20:59:04	Araq	these are abstract symbols I doubt my vision has anything to do with it
20:59:11	dom96	+ is just weird here
20:59:12	FromGitter	<RedBeard0531> It is a very tall and depressed +
20:59:23	FromGitter	<RedBeard0531> hanging its head in shame
20:59:28	dom96	please please reconsider
21:00:18	Araq	RedBeard0531: You can't fix \" for 'fmt'
21:00:24	miran	if only my iterator syntax proposal got 1/10 of the attention this fmt thing has got.... :(
21:00:43	Araq	miran, I read it and I like it. what else is there to say?
21:01:19	miran	Araq: that's nice to know, thanks! i was hoping if somebody else have something to add/say/improve....
21:01:41	*	nsf quit (Quit: WeeChat 2.0.1)
21:01:42	dom96	miran: Is there an RFC for this?
21:01:59	FromGitter	<data-man> StrFormat -> sf"{is}{the}{best}"
21:02:11	miran	https://github.com/nim-lang/Nim/issues/7047
21:02:25	Araq	dom96, what if we stick to % then and live with the incompatibility with json?
21:03:44	*	MJCaley joined #nim
21:04:06	dom96	argh
21:04:21	dom96	what if we provide % and format?
21:04:39	enthus1ast	format formats
21:04:42	dom96	put the % alias into a separate module, like: import strformat.ops
21:05:02	Araq	'fmt' does not work well, again.
21:05:15	Araq	only an operator does.
21:05:25	dom96	it does if you write fmt("..."), no?
21:05:29	miran	can json's % be changed? and why not? :)
21:05:38	dom96	miran: because it's canon
21:05:54	dom96	(my book shows many examples with it)
21:06:00	enthus1ast	§ ?
21:06:19	Araq	dom96, true but that's too verbose for some people
21:06:28	dom96	I feel like there should be a wiki article titled "Nim's historic moments"
21:06:35	dom96	Where we put links to discussions like this
21:06:51	miran	can't you just have errata repo where will be errors plus differences in newer nim versions?
21:07:08	dom96	I can, but I don't want to
21:07:18	miran	i know you're trying to avoid any thing that breaks the code in the book, but....
21:07:20	dom96	Me and Araq agreed that we will release v1 that's compatible with my book
21:07:44	Araq	what's wrong with + ? use it for one day and then come back and complain.
21:08:03	dom96	+ is commonly used for addition
21:08:09	dom96	it's odd to use it for formatting
21:08:12	Araq	unary + is a nop.
21:08:21	Araq	and has no meaning for string literals
21:08:29	dom96	yes, but it has meaning for humans
21:08:57	Araq	I'm a human and it has no meaning to me.
21:08:58	FromGitter	<data-man> wtf"{Oh}"
21:09:10	miran	! in vim is 'execute command', i see some connection with 'evaluate string' :)
21:09:26	FromGitter	<RedBeard0531> Is the only problem with fmt that you need to use """ if you want to embed a " in a {} block?
21:09:27	Araq	pegs use unary + for "one or more"
21:09:31	dom96	yeah, I would prefer ! to +
21:09:56	dom96	Yeah, I'm still unclear about what the problem with fmt"" is
21:10:06	FromGitter	<RedBeard0531> assuming that fmt called newLit(unescape(str))
21:10:20	Araq	fmt"\""
21:12:07	dom96	elaborate
21:12:37	FromGitter	<GULPF> second " ends the string since it's a raw string
21:13:29	dom96	no, that raw string isn't finished
21:13:31	FromGitter	<RedBeard0531> Its a bit odd that you can't \" inside a raw string, but I guess that ship sailed long ago
21:14:17	Araq	fmt"\"" doesn't work
21:14:36	Araq	\/"\"" does
21:15:09	FromGitter	<RedBeard0531> We *could* make fmt require nnkTripleStrLit, but I guess that won't make anyone happy
21:15:23	FromGitter	<RedBeard0531> That is a cut the baby in half solution
21:15:27	FromGitter	<GULPF> @dom96 oh, I didn't realize that `""` is an escaped `"` in raw strings
21:16:09	FromGitter	<RedBeard0531> Yeah, that felt weird to me, but it is what it is
21:16:16	dom96	yep
21:16:27	Araq	doubling the escape hatch is the only way that truely works
21:16:33	Araq	see \\ for an example.
21:16:51	dom96	fmt"{\"}" would be a valid workaround
21:17:23	Araq	so I don't really think these criticisms against "" or '' are valid
21:18:09	Araq	or C's format strings where a percent is written as %%
21:18:46	Araq	by your logic that should have been \% and printf should do the interpretation of \n
21:18:56	alexday	https://nim-lang.org/docs/tut1.html#procedures-result-variable why is the design done like this. the result variable
21:19:00	alexday	implicit
21:19:00	FromGitter	<RedBeard0531> but \ is the escape in strings, not "
21:19:22	FromGitter	<GULPF> but not in raw strings
21:19:29	FromGitter	<RedBeard0531> alexday that has become one of my favorite nim features. give it a shot. You'll probably like it!
21:19:44	dom96	RedBeard0531: I love that feature too :D
21:19:48	Araq	the point of raw strings is that the backslash is untouched
21:19:56	FromGitter	<GULPF> what if you actually want `\"`
21:20:23	FromGitter	<GULPF> `\\"`? then what if *thats* what you actually want
21:20:56	alexday	it sounds good, its unique. but then unless one knows the feature its difficult to know :P
21:21:10	Araq	alexday, Delphi and Eiffel have the feature too :P
21:21:18	Araq	and I took it from them.
21:21:25	miran	alexday: that's true about, well... everything, no?
21:21:48	alexday	oh ok. I see.
21:22:09	FromGitter	<GULPF> I'm starting to think `fmt""` is the best option... IMO backslashes needs to be handled differently inside and outside the `{}` blocks. The wart of having to use `"""` for `"` is OK
21:23:20	dom96	now that I've seen the dreaded +, I will happily take `fmt""`
21:23:28	dom96	Don't even care about f"" anymore
21:23:37	FromGitter	<GULPF> fmt"{\foo}\n" should work, where `\` is an operator
21:23:38	miran	echo "{"a\nb"}\n{"c\nd"}" == ?
21:24:06	miran	with fmt/+/?/whatever in front
21:25:24	dom96	!eval echo r"{"a\nb"}\n{"c\nd"}"
21:25:25	NimBot	Compile failed: in.nim(1, 11) Error: undeclared identifier: '\'
21:25:44	dom96	perfect for this occassion
21:26:08	Araq	ok, one last idea
21:26:22	Araq	we will have both fmt and %
21:26:35	dom96	I suggested that above
21:26:37	Araq	import strformat except % can be used with JSON
21:26:51	dom96	but also putting `%` into a separate module
21:27:06	Araq	why? people need to learn about import except
21:27:07	dom96	i.e. import strformat.ops
21:27:08	miran	can't %/fmt be used without any import?
21:28:24	Araq	strformat.ops is not how anything else in the stdlib does things
21:29:57	*	Trustable quit (Remote host closed the connection)
21:29:59	dom96	I suppose it makes sense to introduce people to this 'except' stuff ASAP
21:30:28	dom96	but I fear it will be too prevalent
21:33:16	*	gokr joined #nim
21:36:17	*	natrys joined #nim
21:38:16	miran	"import strformat except % " --> so this is in separate module from strutils?
21:38:46	FromGitter	<GULPF> https://github.com/nim-lang/Nim/blob/devel/lib/pure/strformat.nim
21:38:50	FromGitter	<GULPF> yes
21:39:26	miran	my imports will be longer than my code :D
21:40:00	Araq	miran, my documentation comments are longer than my code
21:40:11	miran	that's ok :)
21:45:02	miran	well guys. i'm off to sleep, can't wait to see in the morning if it is still %/fmt, something else that already was a candidate, or something completely new :)
21:46:42	*	miran quit (Quit: Konversation terminated!)
21:49:38	*	Jesin quit (Ping timeout: 255 seconds)
21:53:32	yglukhov	Imo, all operators should be lightweight wrappers to something with a more meaningful name
21:53:48	dom96	+1
21:54:28	yglukhov	especially when those operators are macros, which are sometimes don't behave well in overloading resolution
21:55:22	yglukhov	e.g. https://github.com/nim-lang/Nim/issues/6643
22:03:04	Araq	that issue uses 'foo' which is not an operator.
22:03:36	Araq	what's the logic here? 'untyped' is a bit aggressive, yes.
22:03:49	Araq	the connection with operators doesn't exist.
22:07:29	*	Ven`` quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
22:11:26	*	MJCaley quit (Quit: MJCaley)
22:15:08	*	Sentreen joined #nim
22:17:29	yglukhov	Araq: the logic is that operators are more likely to clash, because everyone thinks that his particular function is so extremely handy that it's gonna be used all over the place, so let's make this `%` = toJson(smth). Why `%`? because arithmetic operators are definitely meaningless here, and would just confuse everyone. So let's take some really meaningless operator and give it some meaning :D
22:18:02	yglukhov	and `%` is the first meaningless operator that comes to mind
22:18:21	yglukhov	just kidding of course, but you get the point ;)
22:20:16	*	joshbaptiste quit (Ping timeout: 268 seconds)
22:22:23	*	natrys quit (Quit: natrys)
22:23:05	Araq	I don't really. `%` doesn't use 'untyped'
22:23:31	Araq	and the stdlib is careful with its operators
22:23:51	Araq	case in point: % used to be fmt until we figured out it sadly doesn't really work well.
22:24:49	*	claudiuinberlin quit (Quit: Textual IRC Client: www.textualapp.com)
22:25:16	*	vlad1777d joined #nim
22:25:18	Araq	where are these operator clashes that you see?
22:26:04	Araq	and how would 'asgnPlus' instead of '+=' help?
22:27:11	yglukhov	% used to be (a) lightweight, non-generic (b) proc.
22:27:23	yglukhov	clashes may occur in non-std libs
22:28:18	Araq	json.`%`(a) # disambiguate. it works.
22:28:40	Araq	I think this is really nicely done in Nim and don't see why we need to change anything
22:29:35	Araq	now ok, json.% in particular it a bit misdesigned as the single %* that was introduced later would do the same job
22:33:22	yglukhov	"I think this is really nicely done in Nim" - i think so too. however, there's a bson module in nimongo, which has introduced `%*` as well (ok, that was a mistake, but still). now any module that imports bson and json (a lot of our server code) is doomed to always do json.`%*`(smth) and bson.`%*`(smth). With my suggestion we could just opt out to toJson(smth) and toBson(smth), which is (a) shorter and (b) meaningful.
22:34:31	Araq	well you can do that with a short helper module
22:34:33	yglukhov	other than that i'm completely happy with using the operators, and i like them. i'm just suggesting a friendly practice
22:35:27	Araq	it's unfortunate but you can put the discipline somewhere else ("import bsonex" to make things easy to write)
22:37:10	yglukhov	Araq: ok, here's another sample. this macro `%*` adds `%` to the ast nodes of its input
22:37:29	yglukhov	it hopes, that `%` will return JsonNode for MyType
22:37:58	yglukhov	now what if bson.`%*` wants the same?
22:38:16	Araq	it doesn't hope, I fixed that long ago, it uses bindSym"%"
22:38:17	yglukhov	and i want MyType to be both json and bson compliant
22:38:31	Araq	so it only uses the %'s in json.nim
22:38:32	yglukhov	even worse
22:38:47	dom96	To all the people coming to FOSDEM, you guys want me to get Nim t-shirts for you?
22:38:57	dom96	If so, tell me your size
22:39:01	yglukhov	so i can't put mytype into the json literal
22:39:06	yglukhov	dom96: yes!
22:39:35	yglukhov	L
22:40:06	dom96	Oh, also need a colour preference:
22:40:12	dom96	Denim Blue: https://uktshirtprinting.com/product/custom-product-5a5673687806c3-14363436
22:40:20	dom96	Dark Grey: https://uktshirtprinting.com/product/custom-product-5a56738b901695-06005377
22:40:27	dom96	Black: https://uktshirtprinting.com/product/custom-product-5a5673baf0ecc8-16150316
22:40:40	dom96	Navy Blue: https://uktshirtprinting.com/product/custom-product-5a567eba2753f5-71020628
22:40:53	Araq	yglukhov, ok but now the problem shifted from "% clashes" to "%* doesn't support a protocol that it otherwise could"
22:41:12	Araq	which is true and we should fix json.nim
22:41:13	dom96	There is also a fabric choice, one is ~£19.00 the other is ~£24.00 (but has less colour choices: just black and grey)
22:41:39	yglukhov	dom96: black lgtm
22:42:02	dom96	The cheaper or more expensive one?
22:42:09	yglukhov	Araq: well, with my friendly practice this would never be a problem
22:43:27	yglukhov	but ok, maybe there's no point in arguing about it until the next nice use-case.
22:43:29	FromGitter	<RedBeard0531> @yglukhov We finally invested some time a few weeks ago to tune our server-internal BSON parser. If you are using bson a lot it may be worth doing the same thing in nimongo: https://github.com/mongodb/mongo/commit/3c586a7c946ac7f9485b9ff978fa029ac4b5108e
22:44:17	yglukhov	dom96: let it be the premial =)
22:44:26	dom96	yglukhov: okay :)
22:45:25	dom96	I wonder if I should ask on the forum
22:45:31	dom96	not much time to order these now though
22:45:43	yglukhov	RedBeard0531: thanks! even though bson performance is not our main problem atm =)
22:47:04	dom96	PMunch, federico3 ^^ any special requests for t-shirts?
22:47:13	*	joshbaptiste joined #nim
22:50:11	*	Jesin joined #nim
22:51:43	*	arecaceae quit (Ping timeout: 268 seconds)
22:51:54	dom96	btw any comments on these designs welcome
22:52:59	*	arecaceae joined #nim
22:54:23	*	devdri quit ()
22:56:53	*	chemist69 quit (Ping timeout: 276 seconds)
23:07:37	*	yglukhov quit (Remote host closed the connection)
23:08:10	*	yglukhov joined #nim
23:09:41	enthus1ast	join c
23:09:56	*	chemist69 joined #nim
23:09:56	enthus1ast	sorry
23:11:00	*	dddddd quit (Remote host closed the connection)
23:11:45	FromGitter	<zetashift> @RedBeard0531 I
23:11:57	FromGitter	<zetashift> ve heard some mixed things about MongoDB, where does it shine?
23:12:27	*	yglukhov quit (Ping timeout: 240 seconds)
23:14:16	FromGitter	<RedBeard0531> I've been working on it for over 8 years now so I'm a bit biased :)
23:16:28	FromGitter	<RedBeard0531> But we try really hard to make it damn easy to use. We focus on dev productivity for people building apps that use the DB. Probably best to DM if you have more questions though.
23:16:47	*	endragor joined #nim
23:17:16	*	MJCaley joined #nim
23:21:03	*	endragor quit (Ping timeout: 248 seconds)
23:26:31	Araq	RedBeard0531: Oh are you working for them?
23:27:11	Araq	interesting, Mongodb saved my butt once
23:29:26	FromGitter	<RedBeard0531> Yup. Glad to hear it! Any chance it was a feature I worked on?
23:36:16	*	MJCaley quit (Quit: MJCaley)
23:39:31	*	yglukhov joined #nim
23:41:40	FromGitter	<zetashift> Well I'm not asking why it's better than others, what kind of problems does MongoDB solve easier than others. I've used it twice before in small js sites and I liked how easy it was to get started with
23:43:22	Araq	RedBeard0531: Windows support? :-)
23:43:53	Araq	performance?
23:44:09	*	yglukhov quit (Ping timeout: 264 seconds)
23:44:11	Araq	being easy to setup and deploy?
23:44:32	Araq	the only downside is your query language :-)
23:45:29	FromGitter	<RedBeard0531> heh. I try to avoid working on our windows support but sometimes I can't help it. I do spend a LOT of time on perf improvements. I'm the goto guy for when things are slower than they should be.
23:45:51	*	zahary quit (Quit: Leaving.)
23:46:10	FromGitter	<RedBeard0531> Have you tried the agg language? I spent a lot of time on that a few years ago, although I've largely handed it off.
23:46:18	Araq	http://www.globalnerdy.com/wordpress/wp-content/uploads/2014/05/how-do-i-query-the-database.jpg
23:46:41	FromGitter	<RedBeard0531> trust me, we hate map/reduce more than you do...
23:46:55	Araq	that comic never gets old lol
23:47:09	FromGitter	<RedBeard0531> the new (well old now) agg language is our replacement for map reduce
23:47:22	Araq	never heard of it
23:47:36	Araq	it's been awhile since I last used Mongo
23:49:55	Araq	did you fix the "mongodb just keeps growing" problems on Windows?
23:49:56	FromGitter	<RedBeard0531> db.postcodes.aggregate([{$match: {state: "NY}}, {$group: {_id: "$city", totalPop: {$sum: "$pop"}, avgPop: {$avg: "$pop"}}}])
23:50:30	Araq	oh yeah, that's not a query language.
23:50:37	FromGitter	<RedBeard0531> probably? I don't remember that one.
23:51:58	Araq	that's "too lazy to write a parser". ;-)
23:52:00	FromGitter	<RedBeard0531> I kinda like the queries-are-data API. It makes it really easy to build them programatically.
23:52:34	Araq	yeah or that.
23:52:36	FromGitter	<RedBeard0531> Haven't I told you that parsing is my least favorite part of programming ;)
23:52:47	Araq	it's an API, not a query language. fail.
23:53:13	Araq	the trick about "adhoc queries" is in the "adhoc" part
23:53:44	Araq	I don't write programs that construct these queries as often as I write queries.
23:53:45	FromGitter	<RedBeard0531> Most users access it programatically. and not from COBAL.
23:58:26	Araq	*shrug* I used Mongo because it works on Windows and had superior performance
23:58:51	Araq	and losing some data wasn't also a problem.
23:59:44	Araq	the "you have no schema" and "your query is a JSON" parts were a pita ;-)