<< 19-01-2014 >>

---

00:01:48	fowl	Araq, perhaps it should depend on how you instantiate it
00:02:20	vbtt	oh i think i get it. iterator() is just a resumable function.
00:02:33	Araq	vbtt: exactly
00:02:37	vbtt	and you can pass in a different value each time.
00:02:43	Araq	yes lol
00:02:45	vbtt	and the iterator can make decisions based on that.
00:02:51	Araq	yes
00:02:55	vbtt	ok cool.
00:02:58	vbtt	so it's by design.
00:03:04	vbtt	and it's useful.
00:03:05	Araq	yes
00:03:15	Varriount	vbtt: Just so you know, in that snippet, 'iterator' is a keyword, not an identifier
00:03:21	vbtt	it's just the name iterator threw me off.
00:03:46	Araq	well indeed that's now a bad name
00:03:58	Araq	happens when languages evolve
00:04:07	vbtt	and yeah i understand how you can save the arg values with a closure if you want.
00:04:24	vbtt	it's ok. as long as the inline and closure iterators have similiar semantics.
00:04:34	Araq	indeed they have
00:04:45	fowl	Araq, i dont see why both ways wouldnt work though, the python usage is going to be common
00:05:02	vbtt	it's only because i expected python semantics that i was confused. every language has different semantics for iterators.
00:05:28	vbtt	fowl: yes, and it's easy to impelemnt the python usage if thats what you want.
00:05:57	Araq	fowl: that looks like it would require yet another special case for overloading resolution
00:06:13	vbtt	ok g2g
00:06:15	vbtt	ttyl.
00:06:29	vbtt	nimrod looks better and better each time.
00:06:40	*	vbtt quit (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
00:06:46	LordAndrew	To use an entire standard library, do I just do "import strutils"?
00:07:06	Varriount	LordAndrew: You mean, a module? Yes.
00:07:07	Araq	LordAndrew: yes
00:07:55	Varriount	LordAndrew: Some information on importing -> http://nimrod-lang.org/manual.html#import-statement
00:08:41	LordAndrew	Aha, I knew I was missing some sort of documentation. Thanks Varriount. :)
00:09:50	Varriount	LordAndrew: Keep in mind that the manual, though a trove of information, is a bit outdated, and doesn't reflect fully all the features of nimrod
00:10:28	Varriount	LordAndrew: A more up-to-date version of the manual is on the github repo, in the documention directories.
00:15:33	Araq	or here: http://build.nimrod-lang.org/docs/index.html
00:16:06	Varriount	Araq: Yes, but iirc, not even that is *exactly* up-to-date
00:16:27	Araq	true :-/
00:17:08	*	xtagon joined #nimrod
00:24:15	Varriount	Hey xtagon
00:24:59	xtagon	Hey Varriount
00:31:07	*	io2 quit ()
00:31:56	NimBot	Araq/Nimrod devel a30b428 Billingsly Wetherfordshire [+0 ±1 -0]: Update endians.nim... 2 more lines
00:31:56	NimBot	Araq/Nimrod devel b2410ab Josh Girvin [+0 ±1 -0]: Koch now uses PATH/nimrod correctly
00:31:56	NimBot	Araq/Nimrod devel 19574a3 Andreas Rumpf [+0 ±1 -0]: Merge pull request #812 from girvo/koch-fix... 2 more lines
00:31:56	NimBot	Araq/Nimrod devel 250249c Andreas Rumpf [+0 ±1 -0]: Merge pull request #775 from fowlmouth/patch-3... 2 more lines
00:31:56	NimBot	4 more commits.
00:32:56	Araq	dom96: it really sucks to look for the clone command line to clone gtk2
00:33:02	Araq	can't babel do that for me?
00:33:42	dom96	no.
00:35:24	dom96	You can get the url easily using babel though.
00:35:35	Araq	how so?
00:35:41	dom96	babel search gtk
00:36:27	Varriount	Hey Araq, are WideCStrings garbage collected?
00:36:36	Araq	Varriount: yep
00:37:48	Varriount	Araq: Ok, so say I make a call to a windows api procedure, defined using the importc pragma, of the form winCall(newWideCString("Hello"))
00:37:58	Araq	yes
00:38:29	Varriount	Araq: I didn't finish my question..?
00:38:44	Araq	yes
00:38:48	Varriount	When will the string be garbage collected?
00:38:56	Araq	later
00:39:18	Varriount	After the actual api call has completed?
00:39:46	Araq	if the API call is blocking, yeah
00:39:52	Araq	usually it is
00:40:14	Araq	otherwise you have a problem
00:40:55	Varriount	Ok, thanks for the info.
00:41:39	Araq	is there a "git unstash" ?
00:41:49	dom96	git stash apply
00:45:15	Araq	projects\gtk2>git push origin master
00:45:16	Araq	fatal: remote error:
00:45:18	Araq	You can't push to git://github.com/nimrod-code/gtk2.git
00:45:19	Araq	Use https://github.com/nimrod-code/gtk2.git
00:45:25	Araq	gah, I never get this
00:45:48	Varriount	Araq: I feel your pain. However, it could be worse.
00:46:24	Varriount	You could be teaching 10 non-computer-science-interested high school students how to use git.
00:47:55	Araq	non-computer-science-interested high school students sounds like girls
00:48:06	Araq	so that might even be fun :P
00:48:22	Varriount	Nope. A mixture
00:48:41	xtagon	I wish they tought me git in high school
00:49:23	dom96	I wish they taught me anything useful relating to computers in high school...
00:49:24	Varriount	I was leader of the programming team for a local robotics team last year. You would think kids into robotics would be receptive to something like git, but you would be wrong.
00:49:55	dom96	Robotics team!? Those kids don't know how lucky they are.
00:55:12	*	dmac joined #nimrod
01:00:11	xtagon	I took computer programming, game development, and robotics at our homeschool group in elementary school. If I had gone to a brick-and-morter public school instead, I would have missed out on all that fun stuff
01:00:31	xtagon	Give a kid a bunch of programmable legos, and he will go far
01:01:48	xtagon	The most my high school had though was a basic C++ class, but thanks to the internet, you can learn anything on your own :)
01:03:05	xtagon	Varriount, what sort of robots did your team build? Sounds like fun.
01:04:56	dom96	Good night
01:05:06	*	dmac quit (Disconnected by services)
01:05:06	*	dmac_ joined #nimrod
01:05:10	Araq	same here, good night
01:05:26	*	dmac joined #nimrod
01:05:26	*	dmac quit (Client Quit)
01:05:41	*	dmac_ quit (Client Quit)
01:05:51	Varriount	xtagon: Varied by year. We were part of a global robotics group, FRC. One year we built a robot meant to play a robot version of basketball, another year, ultimate frisbee
01:05:59	*	dmac joined #nimrod
01:07:31	Varriount	Huzzah! I now have a pragma cabable of generating helper procedures for all the junk in winlean.nim
01:11:11	Araq	congrats and bye
01:38:31	*	ddl_smurf_ joined #nimrod
01:41:19	*	ddl_smurf quit (Ping timeout: 260 seconds)
01:41:20	*	ddl_smurf_ is now known as ddl_smurf
02:07:19	LordAndrew	Hrm. The split string functions only seem to work with chars?
02:08:04	Varriount	LordAndrew: By chars, you mean, not unicode characters?
02:10:43	Varriount	Or do you mean, not whole strings?
02:10:50	LordAndrew	Not whole strings.
02:11:22	LordAndrew	I wanted to split one large string into multiple strings and store them in a sequence, by splitting them with "<br>".
02:12:23	Varriount	Huh, I guess there isn't.
02:13:28	Varriount	LordAndrew: Lemme check the most recent documentation.
02:15:22	Varriount	LordAndrew: Odd that there isn't an overload of split that accepts strings.
02:15:36	Varriount	LordAndrew: It would be nice if you could make a procedure which does do that. We could then put it into strutils
02:18:30	LordAndrew	I'll see if I can cobble one together. Still learning how to use Nimrod.
02:19:40	Varriount	LordAndrew: Well, I would look at the implementation of split in strutils
02:34:38	Varriount	LordAndrew: You still there?
02:34:44	LordAndrew	Yes.
02:35:34	Varriount	If you aren't super-paranoid about speed, you can use a combination of string slicing and strutils.find/strutils.rfind
02:36:37	Varriount	Find the first occurance of your word, split the string at that occurance, then repeat with what is left
02:36:55	LordAndrew	I'll give that a try. I was looking at how split is implemented and it looks kind of simple, but I'm not sure what the * in the proc name is.
02:37:00	LordAndrew	proc split*(etc)
02:37:09	Varriount	LordAndrew: That is a visibility marker
02:37:23	*	ddl_smurf quit (Quit: ddl_smurf)
02:37:25	Varriount	Any procedure name prefixed by * is public to outside modules
02:37:34	LordAndrew	Ah.
02:37:48	Varriount	Any other questions?
02:38:14	LordAndrew	I think I've got it. I'm glad I didn't assume it was something like a pointer.
02:41:10	Varriount	LordAndrew: Also, something to keep in mind, most things in nimrod that deal with 'ranges' of things are inclusive.
02:44:07	Varriount	So if you do "hello world"[3..9] you get character's 3-9
02:53:39	LordAndrew	How do you insert something into a sequence? thisSeq += "something" throws an error.
02:54:11	Varriount	LordAndrew: Insert, or append?
02:55:37	Varriount	To insert something, use the 'insert' procedure. To append something, use the '&' operator or 'add' procedure.
02:58:44	*	DAddYE joined #nimrod
03:01:16	*	Mordecai joined #nimrod
03:01:55	LordAndrew	Hrm. I seem to be doing something wrong. SIGSEGV: Illegal storage access. (Attempt to read from nil?)
03:02:07	*	psquid quit (Ping timeout: 272 seconds)
03:02:36	Varriount	LordAndrew: Can you give me a gist of the code?
03:03:27	Varriount	Btw, the message you got is the nimrod equivalent of accessing a null value in c, a null pointer error in java, etc
03:05:23	LordAndrew	https://gist.github.com/Lordovos/4b934a4d1e272b0e6386
03:09:05	Varriount	LordAndrew: You didn't initialize 'lines'
03:09:25	*	DAddYE quit ()
03:09:33	Varriount	To create a new sequence, either assign '@[]' to a previously declared sequence type
03:10:06	Varriount	Assign @[item] to a var for type inference
03:10:21	Varriount	or use the newSeq proc (it's a generic)
03:11:32	Varriount	Personally, I would do 'var lines = newSeq[string]()
03:12:58	*	LordAndrew quit (Ping timeout: 272 seconds)
03:19:47	*	LordAndrew joined #nimrod
03:20:16	Varriount	LordAndrew: DId you get my reply about your error?
03:20:38	LordAndrew	I saw the part about not having initialized the array before my Internet connection died.
03:22:34	Varriount	To create a new sequence, either assign '@[]' to a var whose type has already been declared, assign '@[item]' to a new var declaration for a sequence of type 'item'
03:23:01	Varriount	or, LordAndrew, use the newSeq generic proc
03:23:20	Varriount	Personally, I would do 'var lines = newSeq[string]()'
03:24:44	Varriount	But you could also do 'var lines: seq[string] = @[]'
03:25:17	LordAndrew	Is there any sort of difference between the two?
03:26:31	Varriount	Not in any way you or I would notice. Using the newSeq proc allows you to pre-allocate a number of (nil) values to the sequence, but that's about it.
03:28:02	Varriount	There also might be some difference on the machine level, but again - not in any way we would notice.
03:42:45	*	jcrubino joined #nimrod
03:43:43	Varriount	LordAndrew: Made any progress?
03:58:57	*	brson quit (Quit: leaving)
04:00:10	*	xenagi joined #nimrod
04:01:37	LordAndrew	Varriount: Almost. There's a weird blank element at the start of the array though. https://gist.github.com/Lordovos/4b934a4d1e272b0e6386
04:03:09	Varriount	LordAndrew: Is this an attempt for a split proc, or something else?
04:03:53	Varriount	Also, you don't have to stringify len(lines), echo will take care of that
04:05:00	LordAndrew	Ah, whoops. Didn't realize that.
04:06:14	LordAndrew	And yes, it's an attempt at a split proc.
04:06:53	Varriount	LordAndrew: You don't have to put the type for every var, just where the compiler isn't smart enough to infer the type (usually for containers or generic types)
04:07:53	Varriount	Sorry if I sound like I'm harping, I just can't tell if somethings part of your preferred style, or out of not knowing.
04:08:33	LordAndrew	I wasn't sure if it knew or not, haha. Some of the examples have the type specified, and others don't. :S
04:09:05	Varriount	Another piece of advice: There is a repr() procedure which acts like the stringify procedures, except that it works for vastly more types (however, it can't be overloaded/overwritten)
04:09:26	LordAndrew	...wait, what the heck? I just noticed it's putting the strings into the array in reverse.
04:09:30	Varriount	If you use repr on a string, you can see the pointer to the string, as well as any control characters
04:09:39	Varriount	LordAndrew: That is because you are using insert
04:09:43	LordAndrew	http://puu.sh/6q0AT.png
04:10:24	Varriount	Try '&' or 'append', they're less expensive, as they don't have to shift blocks of memory
04:10:54	Varriount	Er, not 'append', 'add'
04:11:41	Varriount	LordAndrew: The empty string is the remainder of what is left after you have found all the occurrences of the seperator.
04:11:57	LordAndrew	Aha. The blank is at the end now with add. Okay, that makes more sense.
04:12:25	*	jcrubino quit (Ping timeout: 272 seconds)
04:12:31	Varriount	LordAndrew: I came across a similar problem in my quick implementation of split()
04:16:16	*	foxcub joined #nimrod
04:18:29	LordAndrew	http://puu.sh/6q12Q.png Strange.
04:18:43	xtagon	In Nimrod is there a way to format strings inline, without the % operator? Like in some languages you can do "Hello, #{person.name}!" instead of "Hello, $#" % [person.name]
04:19:08	xtagon	Can it be done with macros/templates?
04:19:27	Varriount	xtagon: It can be done with macros/templates
04:19:46	Varriount	xtagon: It may be done already. I'll look.
04:19:51	xtagon	Thanks :)
04:20:30	Varriount	LordAndrew: I can't remember if len takes into account the null char at the end or not :/
04:21:01	xtagon	I'm also wondering...is there any way to do namespacing, or are modules always on the top level? How does one separate third-party dependencies from project-specific imports?
04:21:31	Varriount	xtagon: You can do namespacing (look at how nimrod's libs are layed out)
04:22:08	Varriount	xtagon: I believe some things must be done using a project specific configuration file. Beyond that, I'm not sure.
04:22:26	Varriount	I have not had much experience with nimrod projects, outside of nimrod itself.
04:23:00	Varriount	I would look on github or bitbucket, or ask again in about 5-6 hours.
04:23:03	xtagon	Oh I see, it looks like you're supposed to do import "foo/bar"
04:23:10	xtagon	Thanks, sorry for all the dumb questions.
04:23:19	Varriount	I don't mind. I enjoy helping
04:24:05	LordAndrew	Hmm. Are strings by default terminated with \0?
04:24:24	Varriount	LordAndrew: I believe string literals are.
04:25:14	Varriount	Again, I don't know if len() takes that into account or not. Chop the end off of a string literal and display it - is it still visibly the same?
04:25:18	foxcub	Varriount: got a minute to look at some code and tell me if there is a bug in Nimrod, or if I'm missing something?
04:25:24	LordAndrew	var a = "a"
04:25:29	LordAndrew	echo(len(a)) # 1
04:25:36	Varriount	foxcub: Yep
04:25:41	foxcub	https://bitbucket.org/mrzv/serialization/src
04:25:47	foxcub	Look in examples/bintree.nim
04:26:06	foxcub	It imports tables, if it doesn't the code gives an error that in archive.nim, hasKey is not found.
04:26:13	foxcub	But archive.nim itself imports tables.
04:26:17	foxcub	What's going on?
04:26:17	*	jcrubino joined #nimrod
04:27:09	Varriount	xtagon: The string formatting you're looking for doesn't appear to be in the stdlib. I can say, however, that it should be quite possible with a macro
04:27:46	Varriount	Hello jcrubino
04:27:58	xtagon	Varriount, I'll let that be an exercise for me then. I need to learn more about macros anyway
04:27:59	jcrubino	hi
04:28:21	xtagon	LordAndrew, see http://nimrod-lang.org/tut1.html#strings regarding strings and \0
04:28:33	Varriount	xtagon: Get ready to get your hands dirty. AST handling can be very finicky.
04:29:18	Varriount	Though, what you're doing sounds simpler than what I just did, which was generating 2 different procs, each slightly different, using a macro pragma
04:29:35	*	vbtt joined #nimrod
04:29:41	xtagon	Different in what mannor?
04:30:03	xtagon	manner*
04:30:26	Varriount	xtagon: https://gist.github.com/Varriount/8500498
04:30:48	Varriount	It'll only run on windows (or possibly wine) just so you know.
04:31:35	Varriount	Man, you guys are unlucky. All the others live in Europe mainly, and are asleep right now.
04:31:44	Varriount	All you have is me. :/
04:32:32	xtagon	Looking at that macro, all I can think is "At least you didn't have to write it in C."
04:34:30	Varriount	xtagon: *shudder*
04:35:09	Varriount	xtagon: Actually, I like to think that it's fairly straightforward. The complicated parts are when I have to cover all the corner cases in the ast
04:35:36	LordAndrew	I wonder if I could get away with just removing the last element if it has a length of zero.
04:35:49	Varriount	LordAndrew: Want to see my implementation?
04:35:56	LordAndrew	Varriount: Sure.
04:36:21	Varriount	foxcub: I think you found a bug in the library. It has to do with template binding and symbols.
04:36:33	LordAndrew	also dangit i keep trying to throw in semi-colons and braces into my code
04:36:40	foxcub	Varriount: explain
04:36:54	foxcub	Also how do I report Nimrod bugs?
04:36:59	foxcub	On Github?
04:37:23	foxcub	I'd like to boil it down to a smaller example, but so far I haven't succeeded.
04:39:29	Varriount	https://gist.github.com/Varriount/8500547
04:39:36	Varriount	LordAndrew: ^
04:40:29	Varriount	foxcub: Lemme track down the precise error.
04:40:38	foxcub	Varriount: thanks
04:41:48	LordAndrew	Varriount: That looks way different from what I came up with. :P
04:41:53	Varriount	xtagon: I like to think that my macro is rather straightforward. The odd parts are where I have to cover the corner cases, such as where a procedures "name" node can either be an identifier node (if it is private) or a postFix node (if it has a '*' after its name)
04:42:01	Varriount	LordAndrew: Not totally.
04:42:34	Varriount	LordAndrew: I just don't accumulate my result in a sequence, and use more variables.
04:43:36	xtagon	Varriount, I see. So macros work directly with the AST? How do templates differ?
04:45:33	Varriount	Templates are more like formatting a string with input
04:45:59	Varriount	except, you are formatting source code.
04:46:31	xtagon	So that's more like C macros?
04:48:00	Varriount	Yes. Except that you get automatic hygienic variables and such
04:51:31	Varriount	foxcub: I think you might have found a wierd error in nimrod's importing mechanism
04:52:20	foxcub	Varriount: I believe you. Perhaps, I ping Araq the next time he's online.
04:54:17	Varriount	foxcub: The problem is that you and he seem to have timezones that don't overlap very well
04:54:37	Varriount	You could always PM him now, or use memoserv
04:54:53	foxcub	I don't know how that works. I'm not so great with IRC.
04:55:01	LordAndrew	What does the yield keyword do in your snippet, Varriount?
04:55:20	Varriount	foxcub: pm's are private messages, like a private channel between two users.
04:56:11	Varriount	foxcub: Memoserv is an irc service (like nickserv and chanserv) which allows you to send and store memos. It's like a really advanced, irc-network-wide bot
04:56:33	Varriount	LordAndrew: 'yield' is a keyword used by iterators to yield a result.
04:56:56	foxcub	Varriount: I see. Ok, if I can't find him tomorrow, I'll send this PM.
04:59:17	Varriount	foxcub: Does test.nim work?
04:59:26	foxcub	Everything works.
04:59:32	foxcub	But not without import tables
04:59:52	Varriount	Well, I salute you nonetheless on some nice code.
05:00:04	foxcub	Varriount: thanks
05:00:23	Varriount	I see that you emulated python's with open() style. :P
05:01:09	foxcub	What do you mean?
05:01:22	Varriount	the withOpen() tempalte
05:01:25	Varriount	*template
05:01:25	foxcub	withFile?
05:01:32	Varriount	Er, yes, that
05:01:32	foxcub	That's out of Nimrod's doc.
05:01:34	foxcub	Not my idea.
05:01:43	Varriount	Oh. *shrug*
05:01:55	foxcub	But it's a nice style, indeed.
05:07:28	Varriount	foxcub: I explicitly changed the erroring line in archive.nim to "tables.hasKey(a.pointers[], ix)"
05:07:39	Varriount	and the error shifted to line 43
05:07:56	foxcub	Oh, interesting.
05:07:58	foxcub	Let me try that.
05:08:46	foxcub	Ah, brilliant.
05:08:50	foxcub	That works. I get no errors.
05:08:56	Varriount	Oh?
05:09:04	foxcub	Ah, spoke too soon.
05:09:05	foxcub	Never mind.
05:09:31	Varriount	foxcub: I think it also might have something to do with the large number of generic procedures that are being used. They might be confusing the compiler about what procedure to use
05:10:20	foxcub	Varriount: but it shouldn't
05:10:30	foxcub	The problem in line 43 is the same.
05:10:38	foxcub	Instead of hasKey, it's with []=
05:10:44	Varriount	I know that, and you know that, but I'm not sure the compiler knows that.
05:10:50	foxcub	;-)
05:10:56	foxcub	We'll see what Araq says.
05:11:56	*	vbtt quit (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
05:11:58	Varriount	If I didn't know better, I'd say that whatever structure holds the list of overloads for particular procedures is losing certain ones.
05:12:15	Varriount	It's very.. irritating.
05:12:59	foxcub	Well, it seems the consensus is that it's a bug, so I just need to report it properly. I wish I had a smaller example that produced it.
05:13:09	*	jcrubino quit (Ping timeout: 248 seconds)
05:13:56	Varriount	foxcub: Well, how many generics and other procedures can be removed?
05:15:31	Varriount	And anyway, your 'test case' isn't all that big. It's about.. 180 lines altogether, and for a bug this obscure, I'm not surprised that it isn't more.
05:15:36	LordAndrew	Hhhuh, so yield is like return where it sends info back to something, but it doesn't stop the procedure.
05:15:39	foxcub	Hm, I don't know. They are kind of interdependent. I tried to create a simple example from scratch, but it wouldn't fail.
05:16:21	Varriount	foxcub: The only advice I can give is to remove things until what you have starts working.
05:16:30	foxcub	;-)
05:16:35	foxcub	Ok
05:16:44	Varriount	I like to call it "The Jenga Method"
05:17:24	foxcub	In any case, thanks for your help. I'll bug Araq with this question.
05:25:53	*	jcrubino joined #nimrod
05:26:12	Varriount	Alright, I'm about to head to bed. LordAndrew, foxcub, xtagon, any questions?
05:26:31	foxcub	Have a good night.
05:26:36	xtagon	Varriount, nope. Good night :)
05:26:47	LordAndrew	I'm all good. Thanks for the help, Varriount.
05:39:25	*	xenagi quit (Quit: Leaving)
06:14:44	*	LordAndrew quit (Ping timeout: 272 seconds)
06:15:39	*	[1]Endy joined #nimrod
06:24:26	*	foxcub quit (Quit: foxcub)
06:29:50	*	foxcub joined #nimrod
06:36:09	*	xtagon quit (Quit: Leaving)
06:49:55	*	Demos joined #nimrod
07:03:13	*	foxcub quit (Quit: foxcub)
07:12:14	*	jcrubino quit (Ping timeout: 264 seconds)
07:22:43	*	delian1 joined #nimrod
07:24:00	*	jcrubino joined #nimrod
07:25:16	*	capisce_ joined #nimrod
07:25:54	*	aftershave quit (Ping timeout: 264 seconds)
07:26:03	*	capisce quit (Ping timeout: 264 seconds)
07:27:55	*	aftershave joined #nimrod
08:11:42	*	Demos quit (Ping timeout: 265 seconds)
08:29:56	*	Varriount quit (Read error: Connection reset by peer)
08:50:23	*	io2 joined #nimrod
09:06:29	*	delian1 quit (Quit: Terminated!)
09:06:44	*	delian1 joined #nimrod
09:07:05	*	delian1 is now known as delian66
09:38:15	*	dmac quit (Ping timeout: 260 seconds)
09:44:26	*	io2 quit ()
09:44:26	*	zahary quit (Read error: Connection reset by peer)
09:44:56	*	zahary joined #nimrod
09:45:17	*	io2 joined #nimrod
09:56:35	*	jcrubino quit (Ping timeout: 272 seconds)
10:01:40	*	[2]Endy joined #nimrod
10:05:14	*	[1]Endy quit (Ping timeout: 252 seconds)
10:15:57	*	io2 quit (Ping timeout: 272 seconds)
10:21:25	Araq	OrionPK: PR against devel?
10:31:51	*	zahary quit (Ping timeout: 260 seconds)
10:40:21	*	zahary joined #nimrod
10:51:26	*	faassen joined #nimrod
10:54:42	*	zahary1 joined #nimrod
10:54:42	*	zahary quit (Read error: Connection reset by peer)
10:56:12	*	odc joined #nimrod
11:01:10	*	faassen quit (Quit: Leaving.)
12:37:31	*	aftersha_ joined #nimrod
12:47:02	*	aftersha_ quit (Quit: Computer has gone to sleep.)
12:47:20	*	aftersha_ joined #nimrod
12:49:20	*	aftersha_ quit (Client Quit)
12:51:09	*	aftersha_ joined #nimrod
12:51:31	*	aftersha_ quit (Client Quit)
13:19:48	*	renesac joined #nimrod
13:22:56	renesac	hi, I started toying with nimrod, and the manual says that narrowing type conversions should be explicit (ok), but don't shows how I make explicit conversions
13:26:19	*	darkf quit (Quit: Leaving)
13:27:15	EXetoC	renesac: targetType(x), or x.targetType (syntactic sugar)
13:32:25	EXetoC	it shouldn't be necessary when the original value is a literal, but it needs to fit of course
13:35:23	renesac	thanks, I was trying to narrow the output of math.random()
13:41:00	renesac	it would be more efficient using casts, but I would have to narrow it down anyway as it seems to provide only 63bits of randomness
13:41:12	renesac	(I'm filling an array)
13:48:54	renesac	by the way, how I make a vector/dinamic array?
13:50:16	EXetoC	http://build.nimrod-lang.org/docs/manual.html#array-and-sequence-types
13:54:36	EXetoC	cast away if you really need to
14:20:49	*	foxcub joined #nimrod
14:23:05	renesac	oh, I missed sequences, thought they were only sequential numbers of a type, and was looking in openarrays
14:23:36	renesac	there should be an example using newseq[type](len)
14:24:15	renesac	well, the documentation needs much more examples... especially in the modules
14:25:17	renesac	and I managed to get an internal error from the compiler w/o stack traceback
14:25:18	renesac	\o/
14:25:23	renesac	it was my fault though
14:28:47	EXetoC	I don't know about that. It should never happen
14:28:52	EXetoC	not passing "-d:release" to koch might help
14:32:29	EXetoC	with stack traces of course
14:37:13	renesac	right, minimum working example:
14:38:04	renesac	let x = seq.high # save it as error.nim
14:38:16	renesac	nimrod c -r error.nim
14:38:47	renesac	should I file a proper bug report?
14:39:34	EXetoC	it compiles on a recent build
14:40:04	renesac	oh, I'm on 0.9.2
14:56:44	EXetoC	it's recommended to use HEAD. a lot has happened since 0.9.2, and many libs aren't compatible with it
14:57:10	EXetoC	was the executable bit removed from the build scripts?
15:03:59	renesac	the documentation on nimrod-lang is syncronized with HEAD or 0.9.2?
15:04:14	Araq	0.9.2
15:04:27	Araq	nimbuild has quite up to date docs
15:05:45	renesac	oh, only for the standard library right?
15:05:57	renesac	the language itself changed much since 0.9.2?
15:07:32	Araq	no, for everything: http://nimbuild.nimrod-lang.org/
15:08:29	Araq	the language itself is changing a bit but we have a clear deprecation path
15:09:39	Araq	which most people here disagree with, they think it's better to break things asap
15:12:02	renesac	oh, the documentation link in http://build.nimrod-lang.org/ leads directly to http://build.nimrod-lang.org/docs/lib.html
15:12:11	Araq	I know
15:12:21	renesac	so I wasn't finding the language manual
15:12:24	Araq	we'll fix that eventually
15:12:28	renesac	right, thanks
15:26:25	Araq	oh btw welcome renesac! :-)
15:26:49	Araq	where are my manners today
15:30:40	*	LordAndrew joined #nimrod
15:31:18	OrionPK	araq thanks for reminding me, I should have a bit of time the morning
15:33:22	renesac	Araq: thanks, I'm just another programmer in the endless quest to find a better C
15:33:24	renesac	;)
15:34:45	Araq	use --os:standalone and you get that better C
15:35:05	Araq	I wanted to blog about how to write a python extension in nimrod, but I never find the time
15:35:27	LordAndrew	Hi there.
15:37:28	Araq	hi LordAndrew
15:37:29	renesac	Araq: I'm not aiming that low, XD
15:38:32	renesac	and about python, you would have to show where nimrod is better than Cython
15:41:05	LordAndrew	I was curious: what are the files in nimcache that seem to be generated?
15:42:28	EXetoC	it's everything the C compiler deals with
15:42:53	EXetoC	the output should give you that information
15:43:10	EXetoC	by default anyway. "gcc -o..." for example
15:43:48	EXetoC	well, C by default, but there are several backends
15:44:26	renesac	btw, is there a nimrod interactive interpreter?
15:46:28	renesac	it would be handy for experimenting and learning like I'm doing now...
15:47:27	Araq	why? Cython has exactly the shit of a type system that I would imagine
15:47:57	LordAndrew	Ah. What does Nimrod's compiler use GCC for anyways?
15:47:58	Araq	plus an ugly syntax
15:48:16	EXetoC	LordAndrew: for compiling the generated C code
15:48:44	EXetoC	invoke "nimrod" and see the command section
15:49:25	renesac	Araq: but it is derived from python, and is becoming standard for python modules that needs more performance
15:50:21	renesac	even if technically inferior, it is probably the main alternative
15:51:07	EXetoC	LordAndrew: "nimrod --advanced" will show you the other available backends, and an asm backend is being worked on
15:51:35	renesac	which asm?
15:51:56	Araq	renesac: we have "nimrod i" for an interactive shell but that's quite unusable for anything
15:54:08	EXetoC	renesac: for quite a few architectures I think
15:54:51	renesac	based on LLVM or something?
15:55:17	NimBot	Araq/Nimrod devel 21aa955 Araq [+0 ±54 -0]: 'nil' as a statement is deprecated, use an empty 'discard' instead
15:55:40	EXetoC	renesac: made from scratch
15:55:44	EXetoC	Araq: right?
15:55:48	Araq	right
15:56:12	Araq	but the C backend is not going anywhere
15:56:51	renesac	hum
15:57:20	LordAndrew	Does GCC just compile the code? Or does it link libgcc/glibc to the executables Nimrod generates as well?
15:58:49	renesac	"nimrod i" is cool, but I couldn't get a 'for' to work
16:00:49	EXetoC	it's pretty unusable atm like he said
16:01:14	renesac	yeah....
16:01:20	Araq	LordAndrew: it links with C's stdlib
16:01:55	*	io2 joined #nimrod
16:10:15	*	[2]Endy quit (Ping timeout: 272 seconds)
16:18:56	*	Varriount joined #nimrod
16:19:08	EXetoC	is child(PXmlNode) not recursive? I'm trying to scrape the internets, but maybe I should stick to python
16:19:20	EXetoC	or I could attempt to wrap lxml or something
16:19:45	Araq	PXmlNodes are recursive ...
16:20:47	Araq	Varriount: the trailing binary zero is not part of the lenght and an implementation detail. you don't have to care about it, it just works
16:20:57	Araq	*length
16:21:12	Varriount	Eh... What?
16:21:20	Varriount	Oh, about the strings. Yeah
16:24:18	Varriount	Araq: Speaking of strings, did you see my implementation of split(s, sep:string) ?
16:24:31	Araq	no
16:24:42	Araq	didn't gradha already provide this one?
16:25:03	Varriount	If he did, it hasn't been merged into any of the branches.
16:27:24	Varriount	Araq: There is no split(s, sep:string) implementation in the master or devel branches.
16:29:12	Varriount	Araq: There's a re.split, but that requires pcre
16:29:35	Varriount	Araq: Here's my implementation. -> https://gist.github.com/Varriount/8500547
16:32:20	Araq	Varriount: substr is slightly more efficient than [..]
16:32:33	Araq	don't yield twice
16:33:05	Araq	find has a "start" argument so no need to slice
16:33:15	Varriount	Ok.
16:33:43	Varriount	I'm surprised that you don't want me to inline the entire find() proc :P
16:49:15	Araq	no, iterators need to be as small as possible as they already are inlined
16:49:22	Araq	but btw
16:49:33	Araq	you should use findAux or whatever I called it
16:49:43	Araq	so that the temporary data structure is reused
16:52:11	OrionPK	araq https://github.com/Araq/Nimrod/pull/822
16:53:08	Araq	OrionPK: irc.nim looks suspicious
16:53:20	Araq	why got the 'ret != 0' removed?
16:53:40	OrionPK	thats not being checked anywhere else
16:53:56	OrionPK	I'll add it back
16:54:02	Araq	ok
16:56:01	OrionPK	k, updated
16:57:44	*	jcrubino joined #nimrod
17:00:31	*	aftersha_ joined #nimrod
17:02:10	NimBot	Araq/Nimrod devel 9885240 Erik O'Leary [+0 ±5 -0]: Corrected prunesocketset and pruneprocessset... 3 more lines
17:02:10	NimBot	Araq/Nimrod devel b0a5bab onionhammer [+0 ±1 -0]: Added ret != 0 back
17:02:10	NimBot	Araq/Nimrod devel 568ea12 Andreas Rumpf [+0 ±5 -0]: Merge pull request #822 from onionhammer/devel... 2 more lines
17:04:19	Araq	please close https://github.com/Araq/Nimrod/pull/655 then
17:05:23	OrionPK	k
17:10:00	*	Mordecai quit (Read error: Operation timed out)
17:10:33	*	Mordecai joined #nimrod
17:13:28	*	ddl_smurf joined #nimrod
17:15:06	*	foxcub quit (Quit: foxcub)
17:17:30	EXetoC	does findAll(PXmlNode) not recurse further when said tag is found?
17:20:29	Varriount	Araq: What exactly is this 'TSkipTable'used in auxFind?
17:24:29	Varriount	It's not initialized with a length, yet it's an array type, and it's used.
17:38:19	Araq	EXetoC: dunno check the source
17:39:20	Araq	Varriount: use it as a black box or check out booyer-moore-algorithm
17:40:09	Varriount	Araq: I tried examining it, but nimrod -i's repr returns nothing
17:40:28	Varriount	Is that just a side effect of using the vm?
17:40:59	Araq	no idea
17:51:09	*	aftersha_ quit (Quit: Computer has gone to sleep.)
17:52:13	*	randallsquared quit (Quit: WeeChat 0.4.0)
17:55:03	*	tdc joined #nimrod
18:06:41	tdc	http://pastebin.com/56DfEL4H
18:06:56	tdc	does anyone have a hint on how I can debug this?
18:08:30	Araq	dom96: can you help tdc?
18:08:55	renesac	EXetoC: I'm still getting an error compiling my code with the latest build from nimbuild
18:09:15	renesac	Nimrod Compiler Version 0.9.3 (2014-01-13) [Linux: i386]
18:09:22	renesac	Error: internal error: value expected, but got a type
18:09:23	renesac	No stack traceback available
18:09:53	tdc	I tried adding a breakpoint to proc recv*() at line 1080. I can see it and enable it in the debugger but the breakpoint isn't hit
18:09:56	EXetoC	well, I passed it a seq, not a type
18:10:26	dom96	Araq: Sounds like he needs your help more than mine.
18:10:38	renesac	oh, the line that gives that error is exaclty "let x = seq.high"
18:11:07	renesac	I know it is an error, but it shouldn't cause the compiler exiting w/o pointing the line
18:11:13	dom96	tdc: The Nimrod debugger may still have some issues I think. We usually debug by using 'echo'.
18:12:02	Araq	renesac: the error message could be better, but it's rather hard to do that
18:12:08	tdc	okay - np I wasn't whether it was the debugger or if I'm missing something. I am very new to Nimrod
18:12:57	dom96	tdc: Which version of Nimrod are you using?
18:13:06	Araq	the debugger works for me but doesn't trace the stdlib
18:13:07	tdc	github master
18:13:18	dom96	hrm, good.
18:13:25	renesac	right, it was sufficient for me because I just modified that and tried to compile
18:14:08	dom96	sockets.nim(1295) doesn't seem to have a call to OSError, odd.
18:15:23	EXetoC	renesac: and seq actually is a sequence?
18:16:14	tdc	It would be 1294 for you, because I added a {.breakpoint.}
18:17:33	renesac	no, I think it is really the type
18:17:50	renesac	in the error.nim file, that is the first and only line
18:18:04	dom96	tdc: I see. Well, it's a strange error.
18:18:32	tdc	recv return < 0 but errno appears to be 0
18:18:42	renesac	the problem is only the compiler exiting w/o giving the line where the error happened
18:18:57	EXetoC	renesac: ok so you don't have a var with that name. it shouldn't work either way
18:19:07	EXetoC	yes there are still plenty of compiler bugs
18:19:34	renesac	yes, as I said on the first time, it was my fault
18:19:47	dom96	tdc: The only explanation I can see is that something is resetting errno. That used to be a bug which I thought was fixed, maybe it's back.
18:20:54	tdc	the reason it doesn't work is because the site is using ssl. It works fine for a regular get
18:22:36	dom96	oh yes. Change the osError to socketError
18:23:24	dom96	It needs the socket so: socket.socketError()
18:25:04	tdc	well - Not enough data on socket makes more sense at least.
18:26:58	tdc	although it shouldn't really fail because I don't get the error when using Python
18:27:19	dom96	I thought that that error can only happen with non-blocking sockets
18:28:51	EXetoC	Araq: no findAll doesn't, but I suppose it should. I could add a parameter or two to control that
18:29:39	Araq	EXetoC: find does i think
18:30:16	dom96	tdc: Yeah, that error doesn't make sense. If there is not enough data then it should wait for more.
18:31:53	tdc	it happens when it tries to receive the first byte. I thought it might use the "when ssl" section of the code but it's before that.
18:32:16	Varriount	tdc: What version of nimrod are you using?
18:32:23	dom96	tdc: Maybe it didn't handshake.
18:32:37	Varriount	The 9.2 release, master branch, or devel branch?
18:32:42	dom96	Varriount: I already asked him: master
18:32:50	Varriount	Oh. Sorry
18:34:07	EXetoC	Araq: do you mean findAll? because I only see find.
18:34:16	EXetoC	I think I had to turn the elif into an if
18:34:24	dom96	tdc: Check the precise error code you get. There are 4 possibilities: SSL_ERROR_WANT_CONNECT, SSL_ERROR_WANT_ACCEPT, SSL_ERROR_WANT_WRITE, SSL_ERROR_WANT_READ
18:35:20	*	foxcub joined #nimrod
18:37:12	tdc	Nimrod Compiler Version 0.9.3 (2014-01-19) [Linux: amd64]
18:38:54	tdc	How can I print the error code? Is it on the socketError?
18:39:51	dom96	line 346
18:39:54	dom96	just echo ret
18:40:33	*	[1]Endy joined #nimrod
18:41:03	tdc	Return-code 2
18:42:15	renesac	'shl' don't work with uints??
18:46:04	Araq	import unsigned
18:47:39	*	foxcub quit (Read error: Connection reset by peer)
18:48:11	*	foxcub joined #nimrod
18:48:27	renesac	oh, thanks
18:49:30	*	dmac joined #nimrod
18:50:13	*	io2 quit (Ping timeout: 272 seconds)
18:53:15	tdc	Is the SSL_read expected to occur in blocking or non-blocking mode?
18:53:57	dom96	there is no expectation
18:55:45	tdc	Sorry if it's a dumb question but does Nimrod not have a default?
18:56:45	dom96	No. You set whether a socket is blocking or non-blocking with the setBlocking proc
18:56:55	dom96	Well, the default is blocking.
18:57:18	tdc	okay, seeing that I am not using setBlocking I will assume it is blocking then
18:58:04	dom96	If you're using httpclient then you wouldn't be using setBlocking manually anyway. Httpclient would do it, but httpclient is certainly doesn't.
18:58:06	dom96	s/is//
18:58:29	tdc	maybe renegotiation is taking place. I tried a similar program in golang and that failed because renegotiation is not supported
18:58:54	dom96	Maybe, there is a function which needs to be called to finalise a handshake.
18:59:01	dom96	It it called in connect IIRC
18:59:07	dom96	Might be useful to look into that.
18:59:18	tdc	and the Openssl docs mention that SSL_read might return SSL_ERROR_WANT_READ if renegotiation takes place
18:59:41	dom96	and what should be done in that case?
19:00:15	tdc	Set SSL_MODE_AUTO_RETRY with SSL_CTX_set_mode
19:00:32	dom96	ahh, do it then.
19:00:34	tdc	I need to go away for a bit but might try this in an hour or so
19:00:43	dom96	Alright, if it works please create a PR.
19:00:51	tdc	Will do. thanks for your help
19:05:47	*	Varriount quit (Ping timeout: 272 seconds)
19:06:48	*	foxcub quit (Quit: foxcub)
19:15:53	*	Varriount joined #nimrod
19:16:19	Araq	what should news.txt contain?
19:16:35	Araq	sockets.select now returns * instead of * ?
19:17:44	Araq	dom96?
19:18:04	dom96	sockets.select now prunes sockets that are **not** ready from the list of sockets given to it.
19:22:43	Araq	excellent, thanks
19:25:33	NimBot	Araq/Nimrod devel 96f7fd9 Araq [+0 ±5 -0]: compiler warns when you use GC'ed memory and '--gc:none'
19:25:33	NimBot	Araq/Nimrod devel 1c489f0 Araq [+0 ±5 -0]: Merge branch 'devel' of https://github.com/Araq/Nimrod into devel
19:27:33	*	aftersha_ joined #nimrod
19:27:52	Varriount	Araq: Has foxcub contacted you about the strange import bug he found?
19:28:13	Araq	sure and I replied
19:28:29	Araq	it's f(a, b) vs a.f(b) in a generic context
19:28:35	Araq	a known bug
19:30:11	*	io2 joined #nimrod
19:33:43	*	Mordecai is now known as psquid
19:34:49	*	dmac quit (Ping timeout: 272 seconds)
19:34:53	*	Mat3 joined #nimrod
19:35:47	Mat3	hi
19:36:10	Varriount	Hey Mat3, how's your secret project?
19:36:47	OrionPK	the same is true for osproc select
19:37:10	Araq	OrionPK: you didnt fix osproc select, did you?
19:39:25	Mat3	Varriount: fine, it's not secret - I'm working beside on the documentation
19:40:25	Varriount	Mat3: It is to me, since I hardly know what you're working on.
19:40:29	Varriount	:P
19:44:13	Araq	see you later guys
19:44:45	Mat3	a Nimrod subset with stand-alone environment for embedded boards (somethink similar to this: http://www.eluaproject.net/)
19:44:54	Mat3	^something
19:45:01	Varriount	Ooh, interesting.
19:45:09	Mat3	hi Araq
19:45:26	Varriount	I wonder if nimrod would be a good replacement as a language for programming FRC robots.
19:45:47	Varriount	Currently only C++, Labview, and Java are really used.
19:49:45	Mat3	you forget Forth and Erlang (only to mention a few other alternatives)
19:50:37	Mat3	anyhow why shouldn't Nimrod be suited for this kind of applications ?
19:50:54	Varriount	I don't know.
19:57:23	*	wat_ joined #nimrod
19:58:43	Varriount	Araq: How do I get the tester to output to something other than a database?
19:58:56	Varriount	Or take a database and output it to json?
20:02:34	Varriount	Nevermind
20:12:36	wat_	hey guys, I've been checking out nimrod and having fun with it, but references confuse me a little bit
20:12:52	wat_	I get that new creates a garbage collected reference, but
20:13:41	wat_	if I make a reference to an integer and pass it to a function expecting an integer I get a type error, how do I dereference the variable to turn the reference into a value?
20:13:57	Varriount	wat_: Can I see your code?
20:14:21	wat_	let me do something simple
20:15:40	Varriount	wat_: Use gist.github.com, if what your are doing is longer than a single line.
20:16:03	*	xtagon joined #nimrod
20:16:13	Mat3	hello xtagon
20:16:26	xtagon	Hey Mat3
20:17:02	Varriount	Hi xtagon, make any progress on that macro?
20:17:16	Mat3	see yozu later guys
20:17:19	Mat3	^you
20:17:20	xtagon	Varriount, haven't given it a go yet. I still need to read the manual on macros
20:17:44	*	Mat3 quit (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.)
20:17:55	Varriount	xtagon: Here's a procedure that might help you: http://nimrod-lang.org/system.html#705
20:17:57	wat_	https://gist.github.com/anonymous/6bb1f9b90cc957480b67
20:19:54	xtagon	Varriount, it says that proc returns something other than the documented return type. How is that possible in a type safe language?
20:20:39	Varriount	xtagon: TObject is a base object type
20:20:58	xtagon	Oh, right.
20:20:59	Varriount	xtagon: Nimrod isn't so typesafe that you can't break out or go around
20:21:10	wat_	I noticed that there were finalizers for gc references, are there finalizers for raw ptrs like destructors / smart pointers in c++?
20:22:02	wat_	I have to run so I won't be able to reply, but nimrod is pretty interesting, keep up the good work!
20:22:03	Varriount	xtagon: Locals can also do that because it is a magic proc, and likely a compile time one as well
20:22:19	*	foxcub joined #nimrod
20:22:28	Varriount	'magic' meaning that it's built into the compiler
20:22:46	xtagon	Ah
20:26:55	Varriount	wat_: When the compiler cannot infer whether you need to dereference a ref, you must use the dereference operator, '[]' next to the reference identifier.
20:27:25	tdc	dom96: fyi - adding SSL_MODE_AUTO_RETRY appears to fix the problem
20:27:46	Varriount	foxcub: Araq told me that he explained your problem?
20:27:51	tdc	I have created an issue on github for it
20:27:53	*	aftersha_ quit (Quit: Computer has gone to sleep.)
20:27:59	xtagon	Varriount, so a macro take in an AST object, and returns a modified AST object, yes?
20:28:16	Varriount	xtagon: Yes
20:28:28	foxcub	Varriount: yeah, he already knew about this bug. He explained the solution you proposed, but also fixing the call to []=
20:28:40	foxcub	Varriount: we were basically on the right track, but got confused by something.
20:28:53	foxcub	Anyway, it was good to get him to look at it.
20:28:55	Varriount	foxcub: I'm not surprised. I get confused all the time.
20:29:33	foxcub	Well, you were exactly on the right track. I don't even remember why we didn't get []= yesterday. I thought I tried exactly what I have there now.
20:31:20	Varriount	xtagon: For your macro, the macro would probably take in an ast, check that the only node in it is a string literal node, turn the string literal node into a string, and parse it
20:32:26	xtagon	Varriount, I'm reading the part about "Term rewriting macros" right now. Is that what I need?
20:33:46	xtagon	I'm a little confused on how the macros are ran. If I write a macro that checks that the only node in it is a string literal node, does the compiler try to run the macro on every expression until it finds one with just a string literal?
20:34:01	Varriount	xtagon: Yes, *if* you want your macro to work globally
20:34:15	xtagon	That sounds expensive, is it?
20:34:21	Varriount	I have no idea.
20:34:35	xtagon	I'll find out :)
20:35:07	xtagon	So in my check, if it is not only a string literal node, what's the proper way to bail out? Return the AST as is?
20:35:12	Varriount	xtagon: Another route to take is to create a macro with a short name (such as 'af' for auto-format) and use it as a string prefix
20:35:44	Varriount	xtagon: Probably. Or error, since then you know something went wrong with your targeting pattern.
20:35:48	xtagon	Varriount, I like that idea better, because then you only use it when you need it.
20:35:56	Varriount	xtagon: I do as well.
20:37:39	renesac	oh, if I declare "var result = x" I mess up with the magic
20:38:05	*	renesac took a long time to understand why it wasn't working
20:38:37	renesac	a note about this should be added to the tutorial or manual
20:39:04	xtagon	Varriount, I can't find in the manual the syntax for making a string prefix macro.
20:39:25	xtagon	I assume it needs a pragma or something
20:39:56	renesac	'proc test(): int = var result = 2' returns 0, but if I take the 'var' off it works
20:40:28	Varriount	xtagon: No pragma needed
20:41:14	Varriount	any procedure/macro/template which takes a string can be used as a prefix (I think)
20:41:40	Varriount	I know that procs can be used, I haven't tried it on macros or templates
21:03:06	*	[1]Endy quit (Ping timeout: 265 seconds)
21:05:07	*	jcrubino quit (Ping timeout: 272 seconds)
21:07:11	xtagon	Varriount, I am having a real hard time writing a basic macro that changes a string to another string literal
21:08:11	xtagon	Varriount, what's the best way to debug the value of and AST node? AFAIK there is no REPL since it is compiled, and I can't just print the repr because it's in a macro, not runtime
21:10:47	*	tdc quit (Quit: Leaving)
21:11:33	*	jcrubino joined #nimrod
21:12:35	*	foxcub quit (Quit: foxcub)
21:19:16	dom96	xtagon: Use treeRepr
21:19:25	EXetoC	bacon
21:19:29	xtagon	dom96, thanks
21:20:25	EXetoC	dom96: are you two part of the dota elite yet?
21:20:40	dom96	EXetoC: We haven't played in a while.
21:26:24	Araq	xtagon: tut2 on master contains an excessive tutorial of how to approach creating macros
21:26:41	Araq	hi wat_ welcome
21:26:45	xtagon	Araq, great I'll check that out.
21:28:29	Araq	Varriount: thinking about it, you only need to modify 'replace' to get the 'split' you want
21:33:45	Araq	ping zielmicha-cloud_
21:34:27	zielmicha-cloud_	Pong
21:34:40	Araq	oh yay
21:34:50	Araq	did I miss a PR for osproc?
21:35:11	zielmicha-cloud_	Yes.
21:35:26	Araq	damn
21:36:14	zielmicha-cloud_	(It probably won't merge now.)
21:38:27	*	foxcub joined #nimrod
21:38:49	Araq	well since it's somewhat risky please do the PR against the devel branch
21:39:03	*	Varriount quit (Ping timeout: 272 seconds)
21:39:06	Araq	"master" is now supposed to be "stable"
21:39:35	Araq	where "stable" looks like a synonym for "dead" but oh well
21:42:16	zielmicha-cloud_	Ok
21:46:01	foxcub	Araq: what's the ETA for user-defined type classes?
21:46:26	*	Varriount joined #nimrod
21:46:37	Araq	that's zahary1's business and we don't have a new ETA for 0.9.4 either
21:47:19	foxcub	Ok, cool. I'll ping him. But are you saying user-defined type classes are due in 0.9.4?
21:48:49	Araq	no, I don't know
21:50:39	Varriount	Araq, dom96: The builder hub is down.
21:51:39	dom96	Fixed
21:55:26	Araq	hmm supporting f g(a, b) is rather simple to do
21:55:34	Araq	I wonder if it's a good idea
21:56:01	Araq	let a = await f(a, b) # sweet, hu?
21:56:40	zahary1	but isn't this already suppoerted?
21:56:59	Araq	I don't think so let me check
21:57:19	*	foxcub quit (Quit: foxcub)
21:57:39	zahary1	I'm using stuff like "accept foo(x, y)" in the test suite
21:57:54	dom96	It's not.
21:58:05	Araq	do you use let a = accept foo(x, y) ?
21:58:08	dom96	Not when you're assigning the result to something.
21:58:19	Araq	nkCommand is currently limited
21:58:29	zahary1	yeah, I see
21:59:03	Araq	proc foo(x: int): int = discard
21:59:05	OrionPK	how come i cant pass this char* from C into my cstring :\
21:59:05	Araq	let x = foo foo(1)
21:59:11	OrionPK	i keep getting a sigsegv
21:59:19	Araq	--> invalid indentation
22:00:19	Araq	OrionPK: my bet: the cause is entirely different
22:00:53	OrionPK	i've got char key[len]
22:01:00	OrionPK	filling it
22:01:02	OrionPK	printf("%s", key) works
22:01:18	OrionPK	then I pass it to append_header(client, key)
22:01:24	OrionPK	append_header is an exportc'd proc
22:01:34	OrionPK	2nd param is a cstring
22:01:44	OrionPK	when I try to echo it, I get a sigsegv
22:03:38	Varriount	zahary1: Did you design the caas code for the compiler?
22:03:50	zahary1	foxcub: I'll try to have a more stable version of the type classes in 0.9.4, but you should know that they are not completely unusable at the moment. if you want to start some kind of library based on them, now may be a good time - you should be able to navigate past the occasional limitations
22:04:31	zahary1	Varriount: I wouldn't use "designed it". I've spent around a month getting it to work to its current broken state
22:05:01	Varriount	zahary1: Any reason why 'resetMemory' isn't called after every command?
22:05:12	zahary1	there are a lot of places in the compiler that are not prepared for perpetual execution, but gradually things will get better
22:06:46	xtagon	Araq, I read the tut2 on master and tried to write a macro that should return the statement "Hello, $#!" % ["World"]
22:06:55	xtagon	Araq, but it doesn't work. Mind taking a look? https://gist.github.com/xtagon/7358e7f963526a0ef954
22:07:13	xtagon	Araq, I get (6, 26) Error: value returned by statement has to be discarded
22:07:23	zahary1	resetMemory is just an internal proc I used when I was chasing the old memory leaks of the GC. it tries to completely eradicate the effects of the previous compilation, in order for the GC to be required to free all the memory (it was failing to do so and the memory kept growing after each compilation)
22:08:43	Varriount	zahary1: Currently, in 'serve' mode, the compiler will partially repeat the output of the first command it was given, until the reset command is given.
22:09:12	zahary1	it's no surprise that resetMemory fixes the problems (it brings the compiler to a completely fresh state), but also it defeats the purpose of the caas
22:09:44	Araq	xtagon: are you on master or on devel?
22:09:46	Varriount	zahary1: Not if you want to avoid spawning a new process for each keystroke in an ide, when using the idetools commands
22:09:59	xtagon	Araq, master
22:10:09	Araq	try devel then please
22:10:13	zahary1	you want to avoid compiling all the code, not just spawning a process
22:11:36	xtagon	Araq, I was mistaken, I'm on devel. git ref 1aa0b8c
22:11:55	Araq	xtagon: bug report please
22:12:14	xtagon	Araq, will do
22:12:22	Araq	but try to add a "discard" somewhere
22:12:45	Varriount	I didn't know a macro could accept an argument type like a string. :/
22:13:28	Araq	you should watch my talk then
22:13:48	Araq	however "s" is not a string within the body of the macro
22:13:59	zahary1	it is now, I've fixed that
22:14:08	Araq	it's still an AST ... what?
22:14:08	zahary1	as long as you use static[string]
22:14:23	Araq	-.-
22:14:28	Araq	that was no bug ...
22:15:05	zahary1	we had this conversation once, most of the uses really need the string, not the AST tree
22:15:13	*	Varriount hands the Sacred Salmon of Slapping to Araq
22:15:24	zahary1	we discussed it back when I implemented the compile-time rope formatting in the compiler
22:15:31	Araq	yeah but I liked the consistency
22:17:01	zahary1	the problem is that the AST is only usable for built-in types
22:17:20	zahary1	when you get static[TFoo], the AST is very inconvenient to navigate
22:17:39	Araq	not really
22:18:01	Araq	you get (1, "ab") for Foo = object a: int; b: string
22:18:30	dom96	Isn't it simple to get the actual string from an AST node?
22:18:52	zahary1	so, now dynamic typing is as nice as static typing?
22:18:52	Araq	yeah it is simple
22:19:05	*	foxcub joined #nimrod
22:19:52	zahary1	and there is hardly point in having the AST if it's going to have the same structure each time (nkStrLit "foo"). it only makes sense to preserve it, if you get the AST of the call-site
22:19:56	xtagon	I'm not sure where I would add "discard"
22:20:33	Araq	xtagon: before every 'add' call
22:20:39	xtagon	Okay
22:21:04	Araq	zahary1: for static[string] you're right but otherwise it's not even guaranteed you get a string literal
22:21:17	xtagon	It didn't make a difference
22:21:24	zahary1	yes, I'm discussing only static[string] here
22:21:45	Araq	oh ok, well static[string] has other problems
22:21:52	zahary1	there is value in getting the AST of the call-site, but that's something I want in other situations
22:22:10	Araq	but automatic extraction of the value is indeed not one of those
22:22:37	zahary1	what are the other problems you have in mind?
22:23:49	Araq	proc foo(x: static[string])
22:24:01	Araq	proc foo(x: string) # overloaded
22:24:22	Araq	foo myVariable # crashes instead of calling foo(x: string)
22:24:36	zahary1	why does it crash?
22:24:57	Araq	because myVariable can't be evaluated at compile time
22:25:31	Araq	maybe I misread your code though
22:25:40	zahary1	yeah, but the old VM was able to handle it, so we just need to provide more graceful fallback for the new vm
22:26:38	zahary1	the old VM returns nil when you try to evalConstExpr on a variable, so the first overload is not matched
22:27:04	Araq	yeah but I wasn't aware that parts of the compiler use that feature :P
22:27:10	zahary1	:)
22:27:18	Araq	and I don't know whether it's a good idea
22:28:07	zahary1	well, there must be a way to tell whether something is evaluatable or not. you have added to your to-do that you want implicitStatic to be the default
22:28:24	Araq	currently we have 'static' and 'const' and f(X) where X is a vector of constants and f has no side-effects as contexts for compile time evaluation
22:29:10	Araq	now we get 'foo X' where some "foo" can be overloaded to trigger a compile time evaluation
22:29:42	zahary1	evaluation of X, right?
22:30:00	Araq	right
22:30:14	zahary1	every param can be attempted for constant folding (that's what I thought you mean by implicitStatic by default)
22:30:43	Araq	implicitStatic is just an extension of constant folding
22:30:52	zahary1	but anyway, there should be a way to tell if something is evaluatable or not. I don't see how you can call that a bad idea :)
22:31:01	Araq	and a rather natural extension
22:31:26	Araq	whereas your way might be better handled with a callsite 'static' but I'm not sure
22:33:08	zahary1	meh, I don't see nothing wrong with it. I even have gone further with the semistatic type. have you noticed it?
22:33:24	zahary1	type semistatic[T] = static[T] | T
22:33:34	Araq	yeah ...
22:33:54	Araq	saw it and decided to ignore it for now ;-)
22:38:04	zahary1	btw, I also plan to experiment with more proper definition of "specificity" for the overloading. I've left some notes in typeRel.
22:38:04	zahary1	static[int] is more specific than int, because every static[int] is also an int, but not every int is a static[int]
22:38:45	Araq	iirc sigmatch is performance sensitive
22:38:59	zahary1	similar logic works for generic types as well. seq[seq[T]] is more specific than seq[T]
22:38:59	zahary1	yes, I know. that's why I said "experiement"
22:39:12	Araq	you can try to build an array of function pointers instead of the initial case statement
22:39:28	Araq	that should speed up things
22:39:39	Araq	CPUs got stranger :P
22:40:21	Varriount	Could any of you spare some time to look at my winApi macro? I would like to make sure that I'm not missing any corner cases and such.
22:40:33	Araq	array of function pointers is now faster than computed goto, I think
22:40:59	zahary1	hrm, that would surprise me
22:42:43	Araq	also you could try to get rid of the nkHiddenStdConv crap in sigmatch
22:43:11	Araq	and the crazy copyTree's that shouldn't be necessary since we keep the original ast anyway
22:43:37	zahary1	what about the crazy idea with default arguments in non-last positions?
22:43:57	Araq	I'm fine with that, but don't build a state machine for that
22:44:22	Araq	don't go crazy and introduce regex-like matching
22:44:28	zahary1	I wasn't planning. it will look at each proc individually
22:44:33	Araq	"hit or miss" should be enough
22:48:40	Araq	zahary1: one problem is that "evaluable at compile time" is not defined and things get hairy with -d:useFFI
22:49:24	Araq	IMHO you want full FFI power in a "static" section but in the other contexts a more conservative evaluation
22:51:29	zahary1	I think most of the users will be happy with the conservative approach when it comes to static params. over time, we will annotate many of the FFI procs like `sqrt` to have noSideEffect and the situation will improve further
22:52:31	zahary1	the way I see, static gives you one step in convenience over sting literals, because it enables things like concatenations, table lookups, etc
22:52:58	*	Icefoz joined #nimrod
22:53:22	zahary1	I don't expect people to call into FFI-heavy things when using APIs expecting static params
22:53:33	*	Icefoz left #nimrod (#nimrod)
22:54:01	Araq	oh btw seq[seq[T]] already matches better than seq[T]
22:54:33	Araq	see complexDisambiguation :-)
22:55:58	xtagon	Araq, the macro thing works if I return expr instead of stmt (doh!) so I closed the bug
22:56:14	Araq	xtagon: ah, makes sense
22:56:28	Araq	but the line number is wrong then, right?
22:57:44	Araq	zahary1: works without constructing a witness, C++ requires witness construction afaik
22:57:49	*	darkf joined #nimrod
22:58:07	xtagon	Araq, yeah, looks like the line number is wrong
22:58:34	Araq	so reopen the bug ;-)
22:58:38	xtagon	Right-o
22:58:57	zahary1	I've already committed some of the code. I'm defining subset relationships between the generic types and define "more specific" as "strict subset"
22:59:26	Araq	that's what C++ does, right?
23:00:48	Araq	the C# guys said they designed overloading resolution with their "components" goal in mind, aka "adding new code shouldn't break old code" but I never checked out how this can possibly work
23:01:53	Araq	... overloading resolution is fundamentally at odds with this design principle
23:05:38	zahary1	yes, it is at odds indeed. in generic code it's important to be able to change previously made selections
23:06:10	zahary1	the C++ rules are similar in spirit, but I didn't bother to look at the implementations in clang/gcc
23:06:57	Araq	I looked at D's implementation years ago and at FPC's
23:07:05	*	foxcub quit (Quit: foxcub)
23:07:11	Araq	don't bother with these ;-)
23:07:20	Araq	but clang might be interesting
23:14:08	zahary1	it's getting late, I'll be off to bed.
23:14:45	Araq	one more thing: are you fine with deprecating the "nil" statement?
23:14:55	zahary1	sure
23:15:48	Araq	ok, good night
23:15:54	zahary1	:) good night
23:20:21	xtagon	Is there a way to generate AST nodes in a macro in a more declarative way? i.e. not using add everywhere
23:20:49	Araq	xtagon: sure thing, there are 3 ways, but 1 of them is broken
23:21:11	Araq	1. use macros.quote but it's broken
23:21:33	Araq	2. use parseExpr/parseStmt and embed the ast in strings
23:21:42	Araq	3. use helper templates and 'getAst'
23:21:49	Araq	I prefer (3)
23:22:15	xtagon	Cool, I'll look into that
23:22:28	xtagon	Where in the docs are the helper templates?
23:23:21	Araq	I think gradha wasn't aware of these when he wrote that part :P
23:23:53	xtagon	Ah
23:26:16	LordAndrew	Is there any sort of agreed upon style convention for Nimrod?
23:27:13	Araq	indent 2 spaces, TypesLikeThis, mostOtherStuffLikeThis
23:27:48	Araq	constants are special in that they can start with a lower or an upper case letter
23:28:18	Araq	old convention uses T for value types and P for pointer-like types but too many people complained
23:29:09	EXetoC	:<
23:29:40	Varriount	Araq: What do you think of window api overload procs that auto convert string parameters to widecstring or cstring parameters depending on useWinUnicode?
23:30:39	Varriount	Araq: Like this -> https://gist.github.com/Varriount/8512434
23:32:10	Araq	Varriount: I don't mind it, but I think you should leave winlean alone, it's just some internal module really
23:32:27	Araq	instead you should split up windows.nim and clean up the cruft
23:32:32	Araq	nah, just kidding
23:32:44	Araq	translate the important headers with c2nim instead
23:32:51	Varriount	Erm. I've never split up windows.nim
23:33:11	Araq	we need better win API wrappers
23:33:33	Araq	windows.nim is old and absurdly large
23:33:49	Araq	I'm quite sure nobody needs even 10% of what's in there
23:33:54	Varriount	Araq: My mean goal is to have automatic conversion, and since you're against a string to WideCString converter.. this is the next best thing I could come up with
23:33:58	Varriount	*main
23:34:09	Varriount	*automatic string conversion
23:34:13	Araq	oh god no
23:34:33	Varriount	Araq: What's so bad about it?
23:35:02	Araq	so to avoid a converter you have 2 procs instead of 1 *everywhere*? I'd rather have the converter then instead
23:35:09	*	foxcub joined #nimrod
23:35:34	Varriount	Araq: Actually, 3. And the procs aren't *there*, they're dynamically generated with a macro
23:35:49	Araq	ah yeah
23:35:58	LordAndrew	automagic!
23:36:04	Araq	well you need this A vs W stuff in importc so a converter doesn't help anyway
23:36:37	Araq	why do we need 3 ?
23:37:36	Varriount	One is the proc without an 'a' or 'w' suffix, and no type conversion
23:37:43	*	jcrubino quit (Ping timeout: 260 seconds)
23:37:55	Varriount	One is the proc *with* an 'a' or 'w' suffix, and no type conversion
23:38:11	Araq	get rid of the suffixes, what's the point?
23:38:15	Varriount	One is the proc *without* an 'a' or 'w' suffix, and type conversion
23:38:39	Araq	2 procs of the same name but one uses A version and the other W version
23:39:57	Varriount	Uh.. Yes?
23:40:25	dom96	The 2 wrapper procs are exactly the same, except for the name?
23:40:26	Araq	no need for 3 procs?
23:40:51	Varriount	Well, they aren't wrappers, they are api imports
23:41:00	Varriount	But yes.
23:42:42	dom96	yeah, you don't need 3 procs
23:43:05	dom96	And converter looks more sane.
23:43:08	Varriount	One moment.
23:43:16	LordAndrew	Is there a procedure or such that will tell you what type something is? like istype(1) # int, etc
23:43:20	dom96	This way you will generate twice as much code.
23:43:47	*	jcrubino joined #nimrod
23:43:51	Araq	dom96: converter can't work though
23:44:07	dom96	LordAndrew: http://build.nimrod-lang.org/docs/manual.html#is-operator
23:44:20	Araq	as a converter can't inject the import: "FooA" vs import: "FooW" stuff, dom96
23:44:27	dom96	LordAndrew: Although you probably want type(1)
23:44:48	Varriount	Araq, dom96, look again at my gist
23:45:50	dom96	Araq: Why are all the importc's importing the W version then in Varriount's gist?
23:46:21	Varriount	dom96: Because I have to have an example prefix. They could also be the "A" versions
23:46:36	*	io2 quit ()
23:46:54	dom96	So there is a define which determines this right?
23:47:41	dom96	Can't you just declare a const importcSuffix = when defined(useWinASCII): "A" else: "W" ?
23:48:11	dom96	And then concatenate the literal and the const?
23:48:27	Varriount	dom96: No, because I can't use that on identifiers, just strings.
23:48:44	Varriount	And that is what my macro does, when calculating the new identifiers.
23:48:58	Varriount	dom96: I updated the gist again
23:48:59	dom96	Oh i see.
23:49:59	renesac	<Araq> I'm quite sure nobody needs even 10% of what's in there <-- the old openoffice problem: nobody uses even 10% of msoffice's features, but everybody uses a different 10%
23:50:09	dom96	Right. Well I still don't understand why you need 3 procs.
23:50:24	Varriount	It gives those who want performance the ability to pass in WideCStrings (Araq) and those who want ease of use auto-conversion (me)
23:50:44	Varriount	dom96: Why? Which proc do you think could go or be replaced?
23:50:55	Araq	renesac: that's only a problem when you have lots of users ;-)
23:51:02	dom96	Varriount: Either findFirstFile or findFirstFileW
23:51:12	dom96	Varriount: Either the first proc or the second
23:51:41	renesac	be optimistic ^^"
23:52:09	dom96	Varriount: Ahh, missed your explanation. That makes sense.
23:52:23	Varriount	dom96: If I pick the former, then you still have to use 'when useWinUnicode: ... else:...' in code using the findFirstFileW or findFirstFileA proc
23:52:37	Varriount	Which kinda defeats the purpose of the conversion
23:52:46	Varriount	Er, the conversion proc.
23:54:02	Varriount	If I pick the latter, then people complain about having no distinction between the two unicode/ansi versions of the procs when needed, and we lose backwards compatibility
23:57:23	Araq	just do it the way I suggested plus have a $? operator
23:57:45	Araq	$? either creates a wide string or a cstring depending on some -d:setting
23:57:53	Araq	problem solved
23:58:28	Araq	plus you can always call some other string constructor instead
23:58:45	Araq	if you don't want the setting to infere
23:58:51	Araq	*interfere
23:59:33	Varriount	Araq: I assume you're talking about the third proc, the conversion procedure generated for unicode api?
23:59:40	Varriount	*for the