<< 21-06-2014 >>
00:10:40flaviuIs there a way to get VLAs in nimrod?
00:14:54*saml_ joined #nimrod
00:25:51Varriountflaviu: Variable Length Arrays?
00:26:19flaviuYes. I want an array that is stack-allocated with a size I specify at runtime
00:52:59*hoverbear joined #nimrod
00:57:19Skrylarflaviu: maybe
00:57:28Skrylarit compiles to C so you'd need a c-compatible way
00:57:39Skrylarwindows has a way of allocating on the stack with one of the *allocs
00:57:49Skrylarit'd have to be a special module
00:58:48dom96Wow, deja vu. I remember somebody else asking exactly the same thing.
01:05:51*brson quit (Ping timeout: 244 seconds)
02:12:07*kemet joined #nimrod
02:12:36*kemet quit (Client Quit)
02:46:13*q66 quit (Quit: Leaving)
03:21:54Varriountflaviu: Why does it have to be on the stack?
03:24:55flaviuBecause I don't want to heap-allocate for what is effectively a glorified integer
04:03:00*flaviu quit (Ping timeout: 240 seconds)
04:19:07*darkfusion quit (Ping timeout: 252 seconds)
04:31:31*darkfusion joined #nimrod
04:43:02*noam quit (Ping timeout: 264 seconds)
04:45:03*saml_ quit (Ping timeout: 240 seconds)
04:47:47*Jehan_ quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
04:53:41*darkfusion quit (Ping timeout: 264 seconds)
05:00:16*darkfusion joined #nimrod
05:34:49*Dingba quit (Ping timeout: 246 seconds)
06:03:09*eigenlicht quit (Ping timeout: 255 seconds)
06:07:01*hoverbear quit ()
07:17:47*gkoller joined #nimrod
07:31:52*btiffin joined #nimrod
07:43:12*gkoller quit (Ping timeout: 276 seconds)
07:47:52btiffinI'm always sad to see good technical pages yanked off wikipedia. I have a 700 ish page COBOL book. GNU Cobol, generates intermediate C, so, it'll play Nimrod. I'm going to try and write up a sample for a SourceForge discussion. Would mentioning Nimrod in a COBOL book (free mind you), grant it a sliver of 'notable', at least enough to protest the delete?
07:54:50btiffinAnd while I'm here... The plan is to integrate a Nimrod subprogram called from GNU Cobol. Compile and link envisioned as cobc -x callnim.cob nimrodian.c
07:55:50btiffinI see --noMain, wondering what to expect for entry point name. (COBOL will want to generate the main).
07:58:00btiffinand ... just saw --mainmodule, so off to explore now.
08:08:39*Trixar_za quit (Ping timeout: 252 seconds)
08:09:01*Trixar_za joined #nimrod
08:14:41*f0wl joined #nimrod
08:44:11*BitPuffin joined #nimrod
08:44:23Skrylarhrm
08:44:52Skrylari might have to brew some coffee later; having to track down if png is segfaulting because of an internal error, because a nimrod interaction, or because of user error
08:55:09*joelmo joined #nimrod
08:57:12joelmohi all, does this discard statement mean anything here https://github.com/Araq/Nimrod/blob/7778e79f24a1da4dccfeacc0e9936b171fd1eb74/tests/async/tasyncudp.nim#L24-L36
09:05:26f0wljoelmo, its a block comment
09:06:09joelmof0wl: ok thanks
09:16:47EXetoCit's just like any other value, but some have suggested to make this an exception
09:17:53*kunev joined #nimrod
09:20:21EXetoCso it's not strictly a comment
09:24:27joelmoi think it looks weird to use a keyword to make a block comment, wouldnt it be better to maybe have something like three ### to start a block
09:24:59EXetoChttp://build.nimrod-lang.org/docs/manual.html#triple-quoted-string-literals
09:27:53EXetoCsomething like that has been suggested. # or 'when false' is fine with me
09:46:22*kunev quit (Quit: leaving)
09:49:29*gsingh93_ quit (Quit: Connection closed for inactivity)
10:17:16*vendethiel quit (Remote host closed the connection)
10:18:49*vendethiel joined #nimrod
10:24:33*BitPuffin quit (Ping timeout: 240 seconds)
10:25:55*vendethiel quit (Ping timeout: 244 seconds)
10:27:03*vendethiel joined #nimrod
10:56:12dom96btiffin: It certainly can't hurt.
10:56:31dom96btiffin: But the wikipedia guys are the best to ask I guess.
11:06:58*noam joined #nimrod
11:44:40*q66 joined #nimrod
11:58:02*saml_ joined #nimrod
12:02:36*EXetoC quit (Quit: WeeChat 0.4.3)
12:02:57*exirctest joined #nimrod
12:12:19*BitPuffin joined #nimrod
12:21:27*untitaker quit (Ping timeout: 245 seconds)
12:26:53*untitaker joined #nimrod
12:41:33*exirctest quit (Remote host closed the connection)
12:44:30*darkf quit (Quit: Leaving)
14:06:02*saml_ quit (Ping timeout: 245 seconds)
14:16:21*BitPuffin quit (Ping timeout: 244 seconds)
14:19:50*BitPuffin joined #nimrod
14:22:36*goobles joined #nimrod
14:25:02*BitPuffin quit (Ping timeout: 264 seconds)
14:25:53*askatasuna joined #nimrod
15:04:18*exirctest joined #nimrod
15:04:58*eximiusw1staken quit (Ping timeout: 240 seconds)
15:18:46*eximiuswastaken joined #nimrod
15:41:28flyxhm. I figure I can't get Nimrod to parse a proc declaration as proc declaration if I substitute "proc" with a macro I define
15:43:18*hoverbear joined #nimrod
15:43:24*hoverbear quit (Max SendQ exceeded)
15:43:55*hoverbear joined #nimrod
15:55:23*hoverbear quit ()
15:58:50*hoverbear joined #nimrod
16:02:35*hoverbear quit (Client Quit)
16:43:55*mal`` quit (Ping timeout: 240 seconds)
16:45:12*musicalchair quit (Ping timeout: 245 seconds)
16:46:03*eximiuswastaken quit (Ping timeout: 272 seconds)
16:48:43joelmois there any difference in performance between the async io interfaces with callback or await/future? or should i prefer to use one over the other in some cases, maybe this is just a matter of taste?
16:52:24*eximiuswastaken joined #nimrod
16:55:54*mal`` joined #nimrod
16:57:29*eximiuswastaken quit (Remote host closed the connection)
17:03:16*superfunc joined #nimrod
17:05:27Araqjoelmo: in theory the callbacks should be a bit faster since the async is built on top of these
17:05:37Araqbut I haven't done any measurements
17:06:14Araqand we definitely aim for optimizing the whole async stack
17:08:24*superfunc quit (Quit: leaving)
17:08:59*superfunc joined #nimrod
17:10:54joelmoAraq: ok thanks
17:31:08*BitPuffin joined #nimrod
17:41:01*springbok quit (Ping timeout: 240 seconds)
17:42:45superfuncDo you guys know if any one has made some libraries to deal with sprite sheets yet? I don't think SDL handles that directly
17:44:39exirctestallegro?
17:46:13*musicalchair joined #nimrod
17:46:20superfuncIt might, though I don't really wanna bring in the whole library for that
17:52:33*Jesin quit (Quit: Leaving)
18:02:54*brson joined #nimrod
18:34:41*brson quit (Quit: leaving)
18:34:49*brson joined #nimrod
18:36:25*brson_ joined #nimrod
18:36:28*brson quit (Client Quit)
18:40:30*brson_ quit (Client Quit)
18:40:37*brson joined #nimrod
18:42:07OrionPKyou could use horde:-)
18:45:05exirctestthere's probably more to pull in then
18:51:46superfuncIt's all good. I'll make some standalone stuff for it
19:10:31*superfunc quit (Ping timeout: 240 seconds)
19:15:50gooblesNIMRODDDDDDDDDD
19:19:12dom96hi goobles
19:20:38gooblesHI dom96
19:27:59Araqhi goobles welcome
19:29:27VarriountMeep
19:30:54dom96Varriount: I have a task for you
19:31:01dom96interested?
19:35:09gooblesbrson is a rust spy
19:35:33Araqwe know
19:46:58gooblesis nimrod your favorite language araq
19:47:22Araqof course
19:47:27gooblesgermany what r u doing
19:48:48Araq2nd match is always a problem for germany :P
20:03:26*exirctest quit (Read error: No route to host)
20:04:26gooblesyou can't overload functions in Rust, and D is abit boring, so Nimrod wins
20:05:46Araqyay
20:06:04Araqand we haven't even shown you our upcoming new concurrency suff ;-)
20:06:11dom96hell yeah Nimrod wins.
20:06:14*exirctest joined #nimrod
20:06:56*vendethiel quit (Quit: q+)
20:08:18gooblesyou have some good ideas in this language, look forward to seeing it
20:08:46Araq1:0 !!!
20:08:58gooblessomeone score?
20:09:04Araqgermany
20:09:16gooblesmy feed doesn't show it hum
20:09:25gooblesi must be be delayed
20:09:41gooblesok they scored
20:09:51gooblesgoood job germany
20:10:04dom96who are they playing against?
20:10:13gooblesghana
20:11:11gooblesnaked guy?
20:11:11goobleswtf
20:12:08Araq1:1 :-(
20:12:12Araqfuck
20:12:21gooblesnoo
20:13:13gooblesghana isn't bad not that suprising, hope germany wins though
20:13:39dom96damn you, not I have to watch it
20:13:42dom96*now
20:13:55dom96Didn't the US win against Ghana?
20:14:03gooblesyea
20:14:16joelmo2 - 1, in that match
20:14:52dom96Surely Germany is much better than the US?
20:15:00gooblesthey are yah
20:15:07gooblesbut seems to be sucking in this game
20:15:13goobleswell not scoring;0
20:15:17joelmobut they are much worse if they loose this game :o
20:15:59gooblesthe US won against ghana but didn't look too hot
20:16:09gooblesmostly defending
20:16:18gooblesand getting injured
20:21:14dom96ooh, another score for Ghana
20:21:24gooblesreally?
20:21:35gooblesshit..
20:22:40joelmojust above the head and between the hands on the goal keeper
20:22:49goobleshum well if the US beats Portugal maybe they will be top in the group hah
20:24:34gooblesgermany clobbered portugal can't belive they are loosing now
20:26:11Araqoh well
20:26:25Araqgermany simply has no balls
20:27:08Araqwhenever the other team scores they act like little girls
20:27:31AraqI wonder if I'll ever see a different team
20:28:43Araqah, I take it back :P
20:28:45Araq2:2
20:28:53dom96lol
20:29:26goobleswas it the kloser
20:29:33gooblesoh ic
20:29:35AraqKlose, yes
20:29:38gooblesyep
20:33:24*vendethiel joined #nimrod
20:39:38Araqargh
20:39:42Araqwhy`????
20:42:09Araqargh ???? again???
20:44:27gooblesUS is going to top the group;0
20:46:26dom96every time the german coach or whatever is shown he's picking his nose lol
20:46:42Araqdude
20:46:48Araqhe's giving instructions this way
20:47:10Araqpicking his nose means "we need another goal"
20:47:15dom96lol
20:47:42dom96I don't think they need to be told that
20:48:14*Jehan_ joined #nimrod
20:49:01dom96Lol, that goal keeper fail
20:49:31Varriountdom96: I'm currently finishing up the filemonitoring module.
20:49:40VarriountIt should be done by the end of the weekend.
20:49:55dom96Varriount: This shouldn't take you long. I need a file locking proc.
20:50:04dom96Which is cross-platform
20:50:20dom96and preferably today :P
20:50:34VarriountEh...
20:50:39Jehan_dom96: Have a look at lfs?
20:50:47VarriountYou only wanted me to do the windows side of things.
20:51:48dom96Varriount: Meh. Fine I can implement it myself.
20:51:56dom96Thought you may find this fun.
20:52:09dom96But if you don't want to that's fine
20:53:28Araqgah
20:53:33Araq2:2 final result
20:53:39gooblesthat dudes bleeding all over
20:53:55joelmothe coach didnt pick his nose enough
20:56:17Araqjoelmo: you're right. see dom96 ? he didn't pick enough
20:58:37Jehan_???
20:58:50dom96lol Jehan_
20:59:03dom96We're talking about the World Cup.
20:59:32Jehan_dom96: I figured as much. Still doesn't make a whole lot of sense. :)
21:00:06dom96Well I noticed that the German coach seems to be picking his nose everytime the camera is on him.
21:01:41Jehan_Aha.
21:01:43*exirctest quit (Remote host closed the connection)
21:07:38*Jesin joined #nimrod
21:10:57*perturbation joined #nimrod
21:21:39Varriountdom96: Should I emulate change notifications for the directory itself on Windows?
21:22:38dom96Varriount: Not sure. Do whatever you think is best.
21:33:58*hoverbear joined #nimrod
21:36:13*flaviu joined #nimrod
21:36:21AraqJehan_: do you feel like implementing the new 'deepCopy' builtin? extra points if it doesn't use a hash table for acyclic data structures
21:36:46*hoverbear quit (Client Quit)
21:37:01Araqfixing lambda lifting takes much longer than I thought and so I'm behind my schedule ...
21:38:09Jehan_Araq: Umm … I'm honestly too swamped with my day job and the work on Talis right now for any major undertakings.
21:38:32Araqpff major undertakings ... it's a finger exercise
21:38:35Jehan_As an aside, acyclic structures may still need lookups. Tree structures don't.
21:38:59Araqoh hmm good point
21:38:59Jehan_Araq: If I were more familiar with the compiler internals, probably. :)
21:39:17VarriountJehan_: Talis?
21:40:01Jehan_Varriount: Programming language I've been working on for the past few years.
21:40:29Jehan_Thought I'd mentioned it before?
21:41:00Araqnope
21:41:22Jehan_Maybe not by name, but some concepts I'm pretty sure I've discussed.
21:41:50Jehan_Different goals than Nimrod (and much, much less maturity), so don't be worried about competition. :)
21:43:31Araqha, I'm never worried about competition
21:43:51Jehan_Heh. So far, it's mostly been a research testbed for concurrency ideas for me.
21:44:33Araqeverybody should know by now that PL succeed when they're native for some platform
21:44:43Araqthere are exceptions to this rule
21:44:45Jehan_I *have* contemplated rewriting the compiler in Nimrod so that I can dump C++, though. :)
21:46:13Araqthat's a good idea but then why did you start in C++? why not Ocaml?
21:46:18Jehan_Araq: That's what I mean by different goals (in part). I have not been writing a C++/Java/whatever competitor, I've been writing something that's primarily for research. Widespread adoption is not a major concern.
21:47:01Jehan_Araq: Number of reasons, portability being one of them, especially since it's generating C++ already.
21:47:31Jehan_Mind you, C++ means a greatly simplified subset with operator new/delete overloaded globally to use the Boehm GC.
21:47:47Araqah ok
21:47:54Araqso C++ with a GC
21:48:21Araqbut still you're left with the most retarted way to do pattern matching
21:48:26Jehan_And minus a lot of annoying garbage.
21:48:47Jehan_Not a feature I really need.
21:49:07Jehan_That I can't do "case 'a'..'z'" for the lexer is a bigger annoyance, to be honest. :)
21:49:22AraqGnu supports that
21:49:54Araqand since gnu supports it, so do LLVM and Intel
21:49:54Jehan_Clang also, but not portable.
21:50:20Araqportable enough. Nimrod happily generates it :-)
21:50:28Jehan_I resolved it by having an array that maps characters to character classes. Sort of how Flex does, minus the automation.
21:50:50Jehan_Huh, seriously? Does Microsoft's C compiler handle it?
21:50:58gooblesno
21:51:06Araqno. we don't generate it for Visual C++
21:51:14Jehan_Gotcha.
21:51:17gooblesmsvc hates me
21:51:38gooblesand i hate it back
21:51:42flaviuAraq: I've noticed that bitsets can get pretty big
21:52:09Jehan_But yeah, I write a C++ that is more like Eiffel with braces.
21:52:10Araqflaviu: we have a check for that, can't get bigger than 16K iirc
21:52:10gooblesthen u got alot of bits
21:52:42AraqJehan_: I smell the visitor pattern ...
21:52:45Jehan_To the point of not having explicit header files but generating .h and .cc files via a script from the same source.
21:53:36Jehan_flaviu: Capped at 2^16 bits last I tested.
21:54:11Jehan_Araq: And why not? :)
21:54:25flaviuStill quite big. I'm sure that a variant of run length encoding would make it much more reasonable
21:54:46Jehan_flaviu: Yeah, but then you'd lose constant lookup time.
21:54:52goobleswhy do you have so many bits
21:55:04Jehan_Better to write a special type for that.
21:55:48flaviuWith caches and everything, I'm sure that a linear lookup time would be faster than constant
21:56:10flaviuEspecially when n is something like < 5
21:56:10Araqrumors exist we have a TSet in the stdlib too
21:56:36Araqthe basic types use a single algorithm, flaviu
21:57:16Jehan_flaviu: Well, as soon as you introduce branches, I wouldn't bet on that.
21:57:33joelmoJehan_: do you have Tails published? I want to look at it
21:57:44Araqwe don't use fancy trees for strings either
21:57:53flaviuHash sets are even worse for large sets.
21:58:09Jehan_joelmo: Not yet. That's part of why I'm going to be busy the next few weeks (worst case, months).
21:58:32Araqthere can be better for n < 5 though, flaviu
21:58:38flaviuJehan_: With n likely being small, there would be exactly 1 branch
21:59:09Araqbut yes. keep changing your requirements and neither TSet nor set is good enough for you
21:59:19flaviuAraq: Sorry, I didn't explain in detail what I mean
21:59:56Araqsqlite is also not good enough when n > 5 billion
22:00:17Araqmaybe we should tell them
22:01:51AraqI had a Judy array/table implemented in Nimrod but lost the code
22:02:12Araqthe code was very simple thanks to nimrod's iterators
22:02:38goobleswhere is my std::vector
22:02:47flaviugoobles: seq
22:02:48Araqthe Judy implemenation is only hard because C sux
22:04:00flaviu{1..20,40..1000} would translate to [0, 1, 20, 19, 960] (implementation could be adjusted for signed integers). First, it reads that 0 elements are part of the set. The next 1 element is not part of the set, but the next 20 are. Next 19 are not, next 960 are.
22:05:42flaviuWhile I haven't fleshed out implementations of other operations, I feel there are probably some really elegant and fast ways of implementing them
22:06:31Araqnot really
22:06:54AraqO(lg N) for 'in' when you keep it sorted
22:07:52flaviuSure, but I can keep a thousand of these things around in L1
22:07:59*superfunc joined #nimrod
22:08:01*perturbation quit (Quit: Leaving)
22:08:11flaviuPlus the overhead of binary search is high
22:08:48gooblesif a == 0 || a >1 && a < 20 etc yahhhhhhh
22:09:20flaviugoobles: Thats an optimization in some cases, but how do I do `or` on that?
22:09:51goobleswhat do u mean or?
22:10:02Araqunion I guess
22:10:08*hoverbear joined #nimrod
22:10:32*hoverbear quit (Client Quit)
22:11:40gooblesso you have two sets and want the union, or cases that are true for both?
22:12:15Araqflaviu: if you keep the set as a function you can JIT things and transform data into code
22:12:28Araqnot sure if that's what goobles meant though
22:12:55flaviugoobles: I mean the union of sets
22:13:31flaviuWithout doing anything clever like modifyin code at runtime
22:13:52*hoverbear joined #nimrod
22:14:04flaviuI guess function pointers would work, but that adds overhead
22:14:10*hoverbear quit (Max SendQ exceeded)
22:14:27gooblesi'm not really sure what u are trying to do, or what this is even for.. but you could just write a for loop that performs the logic you mentioned with no need for a giant bitset
22:15:00*hoverbear joined #nimrod
22:15:19Araqflaviu: sets in nimrod are used for:
22:15:28Araq* set[char]
22:15:38Araq* set[someEnum]
22:16:02Araqfor both the existing approach via a lookup table is the most efficient way to do it
22:16:23Araq{0, 1, 4000..8000} is not a realistic example
22:16:47Araqthough you have point for set[UnicodeChar]
22:16:58Araq(which we don't support)
22:18:16Araqand your linear search is a good optimization for TIntSet which is used everywhere in the compiler
22:18:40AraqI never found the time to implement this optimization
22:18:44flaviuAraq: I'm not trying to convince you, really just brainstorming. You're right that the current implementation is best for 95% of usecases
22:19:02Araqhowever, for these the run length encoding doesn't buy you anything
22:19:21Araqas the integers these store are more random and ranges only exist by chance
22:22:08gooblesi tried const fact = fact(some number) to see how compile time evaluation works, it seems pretty slow? Also it explodes if I make the number very large
22:22:19gooblesthough that might just be overflow;0
22:22:26gooblesidk
22:22:38*hoverbear quit (Ping timeout: 264 seconds)
22:22:49Araqshould be pretty fast, in fact
22:23:06Araqdo you use 0.9.4?
22:23:43Araqalso overflow should produce an overflow exception
22:23:49Araqand not an explosion
22:24:24gooblesoh ur right it says overflow or underflow
22:24:31gooblesa nice explosion
22:27:00flaviugoobles: I'm no where near as familiar with the VM as araq, but it should be pretty fast. Are you using recursion? I'm not really sure how it behaves
22:27:48flaviuHe'll probably jump in about now and tell me that recursion is very optimized, I'm really not that familiar with the VM
22:28:08gooblesactually it is fast now;0 I was confused why I couldn't divide an integer by 50
22:28:30Araqrecursion is not super fast, but function calls are as efficient as they can get
22:29:03gooblesit compiles my tiny program in 0.268 seconds not bad
22:29:59flaviugoobles: Since you seem new, I have http://nimrod-by-example.github.io partially completed, read it if you want, and I'd appreciate any feedback possible (especially omissions that have bit you)
22:30:51Araqoh also our VM is "stack less" lol
22:31:07Araqno idea how this can even be considered a feature tbh
22:31:25gooblesi'll check it out flav
22:31:42gooblescan i make a proc without having to specify the types it wants(like generic)?
22:31:54goobleslike use var as the arguement
22:32:04Araqusing the C/hw stack for the interpreter's own call stack like Python does it is ridiculous
22:32:11flaviugoobles: Sure, `proc foo[T](a: T): T`
22:32:17Araqgoobles: yes, it's an implicit generic then
22:32:25Araqpro foo(a) = ...
22:32:30Araq*proc
22:32:56Araqwe got this feature from Clay
22:33:04AraqI'm still not sure I like it
22:34:07goobleskeeps saying invalid indentation but it looks the same as before
22:36:34Jehan_Araq: Huh, interesting, I hadn't realized that actually worked.
22:37:41Jehan_I'm still in favor of declaring the type.
22:37:50gooblesaraq when I use the generic proc you showed it says "undeclared indentifier: result"
22:37:57Jehan_Subjective, of course, but it makes things look more consistent.
22:38:10flaviugoobles: Did you forget a return type?
22:38:25Araqgoobles: well you can't leave out the return type
22:38:39Araqwith a bit of luck 'auto' works as return type though
22:39:52AraqJehan_: leaving out the param types would be sweet for the compiler itself but I don't want it to be generic then, I want ML style type inference for these parameters
22:40:11superfuncgoobles: welcome
22:40:15Jehan_Araq: I see.
22:40:35gooblesAraq: you mean like : proc fact(x) = 'auto'?
22:40:37Jehan_I'm not a big fan of type inference for function arguments and return values myself.
22:40:40gooblesor proc fact(x) = auto
22:40:50flaviugoobles: proc fact(x): auto
22:41:04flaviuproc fact(x): auto = ???
22:41:05Jehan_Part of the value of types is that they are documentation.
22:41:16gooblesok thanks
22:41:55dom96What do you guys prefer? remove, uninstall or delete for the command used to remove packages?
22:42:03Jehan_Which is why I always specify the return type of a Scala function, even though the compiler let's me omit it.
22:42:23goobleseven C++ lets you avoid specifying types nowadays
22:42:28superfuncI've always been a fan of purge from debian
22:42:36gooblesfor some stuff at least
22:42:43superfuncgoobles: C++ does it because its types are f'ing hideous
22:43:03Jehan_goobles: That's something different, that's not for an externally visible API.
22:43:06gooblesya thats why you don't wana look at them;0
22:43:17AraqJehan_: I think we need to overcome this whole idea. readability is at odds with readability
22:43:34Jehan_Araq: Umm?
22:43:46gooblestoo much readability?
22:43:52Araqyou need at least 2 *views* for source code
22:43:52dom96superfunc: That's something I didn't consider, but I don't really like it.
22:44:02Jehan_Readability is at odds with readability? I think you meant a different noun in one case.
22:44:12Araqno
22:44:26Araqthere is the algorithmic aspect
22:44:37Araqwhich is my major interest so I want to read
22:44:39Jehan_Araq: So you want a tool to generate the interface from the source?
22:44:45Araqa[i] += 1
22:45:03Araqbut *many* want the other view
22:45:40Araqcollections.tables.put(a, i, collections.tables.get(a, i) + 1)
22:45:44dom96I'll go for uninstall because it goes well with 'install'
22:46:16Araqthis is the essence of "overloading is bad" etc. which you can find everywhere
22:46:33flaviuAraq: I'd say that there is a balance in terms of readability and verbosity.
22:47:02Araqflaviu: yes, but I think people really want these 2 views
22:47:15superfuncdom96: that seems reasonable
22:47:26Araqthe 2nd view is also called the "confused" view by me
22:47:28Jehan_Araq: Hmm, that's not a version of "overloading is bad" that I'm familiar with.
22:47:49Araqfor the people who are constantly confused when programming :P
22:48:22AraqJehan_: it looks like writability vs readability but really isn't
22:49:13AraqI don't want to *read* collections.tables.put
22:49:19gooblesthat giant ass thing with collections.tables is not code i want to see;0
22:49:30flaviuAraq: But on the other hand, I'm not sure even you want to see code as concise as some regexes
22:49:46Araqflaviu: been there done that
22:49:57gooblesIDE helps alot, you just hover over any concise stuff to get more description
22:50:02Araqadded some whitespace and after a while it was fine
22:50:20flaviugoobles: Not everyone has a fancy IDE. Also, fancy IDEs are lots of work
22:50:52Jehan_One could also argue that IDEs cover up shortcomings in language design.
22:51:01flaviuAraq: My experiences are similar in regards with whitespace, but I still wouldn't say that its especially readable.
22:51:06Jehan_… can cover up ...
22:51:10gooblesthey boost a shitty language into a not so shitty one
22:51:23gooblesbut also make a good language into a great one
22:51:41Jehan_Hmm. I find that I'm generally more productive without an IDE.
22:52:04Jehan_The biggest problem I have with IDEs is that they force me to use their internal editor.
22:52:36AraqIDEs can be superb but most are crap
22:52:41flaviuJehan_: Then you've never used a truly great IDE. Try IntelliJ IDEA
22:53:05Araqflaviu: lol
22:53:11gooblesmost are just terrible text editors
22:53:20Araqgood luck on selling us a Java IDE
22:53:34Jehan_flaviu: Well, the thing is that if an IDE lets me use Vim or Emacs or Textmate or ST or whatever, then I can't access half the productivity features of the IDE.
22:54:03gooblesyou can a file in both the IDE and VIM
22:54:03Jehan_And if you're trying to say that the IDEA editor is good: Umm, no. Just no. It's okayish, but not good.
22:54:31flaviuThe editor is Ok, the vim plugin is Ok, but the tooling it brings is amazing
22:54:34Araqwell an IDE agnostic approach that nimrod idetools tries to provide is surely the way to do it
22:54:49Jehan_Araq: I'd agree with that, at least in principle.
22:54:56flaviuAraq: Yes. And with more emphasis on source-to-source transformations
22:55:11Jehan_The caveat because I'm honestly not sure what the best option is.
22:55:11gooblesidetools looks cool i hope it gets lots of polish and makes for easy IDE's
22:55:56Araqgoobles: be our guest and start working on it ;-)
22:56:14Araqwe're underpowered. Sad but true.
22:56:18gooblesi might at some point once i figure out the basics of nimrod
22:56:28flaviuUnrelated, but I'd like to get feedback on an idea (not a feature request): Structural type inference. If you use methods x(), y(), and z(), then anything that is passed in must have those methods available. It'd allow dynamic looking programs, with static type checking.
22:56:39Jehan_Araq: You need to find a nice spider to bite you. :)
22:57:42Jehan_flaviu: Structural typing exists?
22:58:05Jehan_Technically even in Nimrod for type parameters.
22:58:15flaviuJehan_: Yes, haxe as one example. It has "structs"
22:58:27flaviuScala does too
22:58:28Jehan_OCaml, Scala, Go ...
22:58:47Jehan_(Talis, too, but that's still at the vaporware stage.)
22:59:03gooblescan i suggest that forgetting to put an = sign in this code "proc fact(x): auto =" give a better error msg than "invalid indentation" :)
22:59:56Araqgoobles: you can but nobody cares. it's a newbie mistake
23:00:03goobles:(
23:00:05Jehan_Hmm, the error message usually means that you forgot a token at the end of the previous line.
23:00:21Araqgoobles: don't give me that look
23:00:23Jehan_Maybe the error message should be adapted to include a hint?
23:00:36gooblesno it really was because of the lack of =
23:00:47Araqbetter error messages simply are not a priority right now
23:00:53flaviugoobles: Its hard to give good error messages
23:01:28Araqand usually the error message are really good already. ymmv ofc
23:01:51flaviuWow, Java is getting specialized generics!
23:01:59flaviuAnd structs!
23:01:59goobleswow?
23:02:09goobleszz
23:02:24*darkf joined #nimrod
23:02:39Araqand they'll suck like C#'s structs suck, i bet
23:03:08flaviuHow do C# structs suck? Besides that unsafe features can break value-typeness
23:03:09gooblesand D's
23:03:32*hoverbear joined #nimrod
23:03:38gooblescompare C# struct to C++ struct;0
23:03:48Araqflaviu: try to put one in a collection and then do ++ on one of its fields
23:04:11joelmodom96: apt-get uses 'remove', brew also, and have an alias for uninstall, does not hurt to have both uninstall and remove? i prefer remove
23:04:18Araqthe lack of &T becomes apparent
23:04:28dom96joelmo: Sure, I can add aliases.
23:04:30Jehan_C++ structs are only classes where members are public by default?
23:04:40Araqdom96: I vote for 'rm'
23:04:47AraqJehan_: correct
23:04:49joelmorm is even nicer
23:04:52flaviuOh, I see. Well, that's a weird decision.
23:05:05superfuncdom96: Would it remove any packages that depend on it?
23:05:13flaviudom96: I vote for doing things like git: allow aliases
23:05:25AraqI wasn't serious about 'rm' ...
23:05:33dom96superfunc: no
23:05:34Araqrm doesn't even have a vowel
23:05:58superfuncdom96: Ok, then uninstall is good.
23:05:58dom96superfunc: If there are packages which depend on the package you are removing it will refuse to remove that package
23:06:17Araqdom96: why not allow all of these?
23:06:23superfuncdom96: That's the only reason I mentioned purge. Could be an interesting option to have.
23:06:34AraqI like it that I can currently guess the proper command with babel
23:06:56flaviudom96: Clone pacman's commandline
23:06:58dom96Yes, well. Removing a package is a destructive operation.
23:07:10dom96I don't feel comfortable having too many aliases for it
23:07:15flaviuActually, just fork pacman completely, its awesome
23:07:19Araqonly computer scientists think this way
23:07:35dom96flaviu: screw pacman
23:07:43dom96flaviu: It's not written in Nimrod and therefore sucks
23:07:49Araq"hell ya, I'll have both remove and delete and they should do different things"
23:07:59joelmodont make me have to do: man babel
23:08:49flaviudom96: Or, just say "did you mean `remove`" when it matches any of a set of possible operations
23:08:55Araquninstall|deinstall|remove|delete|del|rm should all work
23:09:13Araqflaviu: see? exactly what I mean
23:09:14gooblesflaviu: i'm looking at the example page you linked me
23:09:24joelmois deinstall english?
23:09:29Araqtrying to "educate" the user with irrelevant crap
23:09:38dom96flaviu: That's just annoying.
23:09:38flaviujoelmo: no, uninstall is
23:09:51gooblesflaviu: maybe mention the generic procs on when you talk about Procedures?
23:10:16gooblesflaviu:the array/seq stuff is blank page
23:10:34Jehan_dom96: Hmm, when I download babel from github it doesn't compile because babelpkg/config is missing?
23:10:49gooblesflaviu: in the procedure it uses result but doesn't explain it
23:10:52dom96Jehan_: oh oops. I guess I forgot to commit it.
23:11:52NimBotnimrod-code/babel master 3f4d8f4 Dominik Picheta [+1 ±0 -0]: Add missing config.nim.
23:12:01dom96Jehan_: There you go.
23:12:10flaviugoobles: several pages aren't completed, but keep the feedback coming. I'll add a link to the result page, thanks
23:12:48Araqremove # simply removes it from Babel's package list
23:12:52Jehan_dom96: Merci!
23:13:01Araqdelete # deletes it from HD completely
23:13:12dom96rm # formats your hard drive
23:13:22Araqpurge # deletes it completely plus all of its dependencies
23:13:51Araqdel # like 'delete' but also sends you an email about what has been deleted
23:14:17Araqrm # like remove but also opens the list of removed files in your $EDITOR
23:14:28Araq<-- Git styled command line
23:14:29dom96I like my rm better
23:14:37Jehan_pdel # Like "del" (the p is silent)
23:14:59flaviuI can't tell if anyones serious
23:15:13Araqit's a good idea if you wanna sell "support"
23:15:25dom96刪除 # same as 'delete' but in chinese
23:15:36Araqgod I can't take these open source zealots seriously
23:15:47Jehan_flaviu: Google: "Leave it to Psmith"
23:17:01flaviuI don't see why people are confused by git, its very simple and elegant. The parameter word choice is sometimes odd, but google is always very helpful.
23:17:05gooblesflaviu: I'm not entirely sure what a "block" statement is for, is it just like a labeled statement ?
23:18:03flaviugoobles: Yes, I'll see if I can make that more clear, thanks.
23:18:59dom96Better yet, i'll use the Google translate API from Babel to allow multi-language commands.
23:19:23Jehan_goobles: block is { … } or begin … end in Pythonesque syntax.
23:19:37Jehan_In other words, a block with its own scope.
23:19:50Araqflaviu: http://stevelosh.com/blog/2013/04/git-koans/
23:20:19gooblesoh ok just a scope gotcha
23:21:46gooblesflaviu: in the result page you have this proc: proc `**`(number, power: int): Is this a function named **? Why does it have string quotes around it? What does that do
23:22:57joelmothose quotes means infix, hakell have them too, means number is lhs, power rhs of expression: number ** power
23:22:58flaviugoobles: The `**` signifies that it is an operator. I should explain that somehow, but I've been having issues coming up with a good order to do things in.
23:22:58Jehan_goobles: The `..` backquotes allow you to treat arbitrary character sequences (well, almost arbitrary) like a regular identifier. It defines an infix operator called **.
23:23:30Jehan_joelmo: No, the backquotes don't mean infix.
23:24:11flaviujoelmo: I can do `@[1, 2, 3]`, the `@` operator isn't infix
23:24:15*hoverbear quit ()
23:24:34Araqflaviu: git lives through the famousness of its creator and its speed. but its interface is just horrible.
23:25:03Jehan_>>> 1.`+`(2)
23:25:04Jehan_3
23:25:08Araqand I'm not sure its speed is really its selling point either ...
23:26:06superfuncwhat is the preffered way to write a while true { ... } block in nimrod
23:26:09Jehan_Araq: Also because there aren't universally superior alternatives (some alternatives are situationally superior, but not universally).
23:26:24Jehan_superfunc: while true: ...
23:26:41dom96Araq: Its selling point right now is Github
23:26:45superfuncThanks Jehan, just making sure there wasn't a keyword I was missing
23:27:12dom96or at least one of the major ones
23:27:14superfuncAraq: I agree, i'd be using hg otherwise
23:27:15flaviuAraq: Some of the criticisms on that page are unfair. a branch is created with `git branch`, a file is reset to the head version with `git reset`, branches are switched with `git checkout`
23:27:25Jehan_superfunc: I wouldn't mind if there were a separate keyword/template for a while true loop myself.
23:27:32AraqI dunno, Jehan_, handling binary files is certainly not a niche problem
23:27:46superfuncI like the rust keyword "loop" for that
23:27:50Jehan_Araq: That's what I mean by situationally superior.
23:27:53Araqgit hardly is a "universal" tool
23:28:27Jehan_Araq: What I'm talking about is that git passes the "good enough" test and there's no tool that beats it on all criteria.
23:28:27Araqsuperfunc: I agree but wanted to save a keyword here
23:28:50gooblesflaviu: it would be nice if it had something on templates, macros etc
23:28:55AraqI don't know the others well enough to be able to say for sure
23:29:02superfuncAraq: Could a macro achieve it?
23:29:08flaviuAraq: Item 3 doesn't really make sense, that's the way things are, and that information isn't particularly useful.
23:29:13Araqsuperfunc: a template even
23:29:16gooblesflaviu: mabye show how to do similiar things in C++ vs nimrod
23:29:22Jehan_Araq: My own preference is Mercurial for most use cases, but I do sometimes want to have something that Git has and Mercurial doesn't.
23:29:25gooblesflaviu: that you might use templates for in C++
23:31:15flaviugoobles: The fact that templates are missing is an issue of non-completion, they are planned. Comparing with C/C++ is a good idea, it can explain things more concisly and with less ambiguity than words. I'll do that in some places, thanks for the suggestion
23:31:41superfuncAraq: When you say you wanted to save a keyword, do you mean to make parsing/compiling less slow/complicated, or to reserve it if its needed in the future?
23:32:26Araqsuperfunc: I simply didn't want to reserve 'loop' for a shortcut of 'while true'
23:33:07Araqparsing it is trivial and doesn't slow down anything
23:34:09Araqnowadays I'd likely do it differently though
23:34:38Araqand have a "loop while" control flow statement
23:35:03Araqloop:
23:35:07Araq foo()
23:35:13Araqwhile cond:
23:35:15Araq bar()
23:36:36Araqor something like that
23:37:01joelmobut you always can put the condition at end with if and break
23:37:11*vendethiel quit (Read error: Connection reset by peer)
23:37:20flaviuAraq: Item 4 is inaccurate, `git tag`, `git remote`, `git branch`, `git status`, 100% consistent. The last two commands are sort of correct, but `git branch -d` has subtly different semantics from `git remote rm` due to how it works.
23:37:20flaviuThe last point is total crap, it says that what is actually a great feature IMO is weird inconstancy
23:37:48superfuncAraq: Interesting. Thanks for the explanation
23:37:57Araqflaviu: "git checkout" already is weird in itself
23:38:32Araqand it goes downhill from there
23:38:36flaviuAraq: It makes perfect sense. It checks out the commit you give it. Branches are just a "pointer" to a commit
23:38:37goobleshg push
23:38:58f0wlrm -rf
23:39:37flaviuI'm not sure where the weird part is
23:39:56Araqflaviu: what if I give a filename? what if I give it nothing?
23:40:00*vendethiel joined #nimrod
23:43:06gooblescan you call C code in a compile time evaluation(const)?
23:43:21flaviuThat's a bit weird, duplicated functionally between `git reset` and `git checkout` is bad. Just don't use git checkout on file names
23:44:07flaviugoobles: No, IIRC it was possible at some point, but I think araq decided it wasn't really worth the time investment
23:46:12flaviuIts in the IRC logs, but I can't be assed to track it down
23:47:21Araqgoobles: you can hack around this limitation via 'gorge' though
23:47:38Araqpeople tend to forget about gorge
23:47:46Araqeven though it's a big enabler
23:47:46Jehan_goobles: the fundamental problem is that at compile time code is evaluated in an interpreted VM, so you have to generate an external program and run that.
23:48:36Jehan_Araq: That's probably because we'd all prefer the name would disappear in favor of staticExec. :)
23:48:58Araqover my dead body :P
23:49:18Araqslurp and gorge give nimrod its soul
23:49:43Jehan_Out of curiosity, where did the names come from?
23:49:54Araqslurp comes from perl
23:50:01Araqgorge is my own addition
23:50:23goobleswhats a gorge?
23:51:13AraqI can't see how slurp and gorge are worse than, say, "fork"
23:51:16flaviuRuns a command and gorges itself on that command's output
23:51:31gooblesoh ic
23:55:14Jehan_Araq: Hmm, is it intended that gorge/staticExec eats all newlines?
23:56:03Araqit should 'strip', Jehan_
23:56:53Araqoh yay
23:56:59Araqit's buggy ...
23:57:20Araqguess that means people only use it for programs that produce 1 line of output
23:57:41Jehan_Araq: Yeah, that was my suspicion as well.
23:57:56Araqwho wants to fix it?
23:58:04Araq1 line of code
23:59:44Jehan_Hmm, what's the semantics that you want?