| 00:10:24 | flaviu | Varriount: Objectively no, but the difference can be felt, and I would say that there should be a prominent button added linking to the build docs. |
| 00:31:14 | * | darkf joined #nimrod |
| 00:37:21 | Varriount | flaviu: Are you looking at the wrapper documentation? |
| 00:37:33 | Skrylar | hrm. siphash can't be used as a stream hash |
| 00:37:39 | Skrylar | it requires the input length ahead of time |
| 00:38:11 | Varriount | Most of those don't have documentation because they are A: Generated by c2nim, and B: Are bare interfaces over whatever API they wrap. |
| 00:38:12 | * | fowl quit (Changing host) |
| 00:38:12 | * | fowl joined #nimrod |
| 00:40:56 | flaviu | You're right, but I wonder if it might be possible to generate documentation too. |
| 00:41:09 | * | carum_ quit (Remote host closed the connection) |
| 00:44:47 | * | io2 quit () |
| 00:46:00 | * | milleral joined #nimrod |
| 00:48:23 | * | Demos quit (Remote host closed the connection) |
| 00:57:21 | Skrylar | flaviu: you could but it would be more work |
| 00:57:34 | Skrylar | at that point you'd basically have to extend c2nim to read doxygen output |
| 00:57:56 | Skrylar | admittedly some people do it; you can ask doxygen to output XML and there are more than a few projects now that use the xml output for scraping metadata |
| 00:59:04 | * | askatasuna joined #nimrod |
| 01:07:27 | Skrylar | Araq, renesac: okay, the siphash port is cleaned up |
| 01:08:19 | * | milleral left #nimrod ("WeeChat 0.3.7") |
| 01:09:11 | * | zahary joined #nimrod |
| 01:09:28 | * | Ifelseif joined #nimrod |
| 01:11:39 | * | carum joined #nimrod |
| 01:17:59 | * | carum quit (Ping timeout: 240 seconds) |
| 01:21:52 | * | Ifelseif quit (Quit: Leaving) |
| 01:25:02 | * | DAddYE quit (Remote host closed the connection) |
| 01:26:10 | * | DAddYE joined #nimrod |
| 01:26:23 | * | DAddYE quit (Remote host closed the connection) |
| 01:26:55 | * | DAddYE joined #nimrod |
| 01:28:00 | * | brson joined #nimrod |
| 01:31:29 | * | DAddYE quit (Ping timeout: 265 seconds) |
| 01:34:15 | * | Demos joined #nimrod |
| 01:39:18 | * | askatasuna quit (Read error: Connection reset by peer) |
| 01:39:54 | * | runvnc joined #nimrod |
| 01:40:21 | runvnc | Hello guys |
| 01:40:38 | runvnc | if you look at https://gist.github.com/ithkuil/9423819#file-gistfile1-nim I am trying to figure out why my sigsegv is not being caught |
| 01:41:06 | runvnc | the process is exiting even though I have a try/except: around the line of code in main that is causing the sigsegv |
| 01:41:15 | runvnc | the exception is in another module. could that be why? |
| 01:41:20 | Demos | sigsegvs are not exceptions |
| 01:41:25 | runvnc | lol |
| 01:41:27 | runvnc | ok |
| 01:41:34 | Demos | if there is a segfault shit has hit the fan and the process will terminate |
| 01:41:56 | Demos | I /think/ you can handle it as a POSIX signal but that is not a very good idea |
| 01:42:15 | Demos | what line is producing the segfault |
| 01:42:24 | runvnc | hm well I think it is quite possible that I could make a mistake that would be a rather occasional problem causing a sigsegv |
| 01:42:31 | runvnc | in other words, this is a basic flaw in the module |
| 01:42:34 | runvnc | that I must correct |
| 01:42:36 | runvnc | but |
| 01:42:39 | * | askatasuna joined #nimrod |
| 01:42:54 | runvnc | I know that I can easily make a similar mistake that will only occur on a more occasional code path |
| 01:43:03 | runvnc | I would like the option for the program to not crash in that case |
| 01:43:22 | runvnc | so the way to do that would be some kind of signal handler? |
| 01:43:28 | Demos | what library is that function from? |
| 01:43:41 | runvnc | is my own module, I know how to correct the segfault by checking for nil |
| 01:44:04 | Demos | and no, not really. a segfault happens when you access memory that you never actually allocated, it is not just accessing a nil pointer |
| 01:44:10 | Demos | then do so |
| 01:45:30 | runvnc | actually I needed to initialize this redis thing |
| 01:45:34 | runvnc | not check for nil |
| 01:46:30 | runvnc | but it still seems easy to have an occasional code path that causes this issue, and I am not sure why it is necessary for the program to always crash in those cases |
| 01:47:55 | Demos | because it means that your state is corrupted, so at best you would get strange behavior further along the line, at worst you would get a security issue |
| 01:48:57 | Demos | if you are using memory and not telling the kernel about it than there is no way anything could work |
| 01:49:37 | flaviu | Do generic immediate templates exist? |
| 01:50:03 | runvnc | I would just prefer that the program would continue at the top of the while loop rather than crashing. of course I don't want that code path to continue with random uninitialized data |
| 01:50:13 | Demos | now you can turn derefs of nil pointers into an exception but it is pointless AND means you have to insert a null check around each deref |
| 01:50:40 | Demos | but what if the segfault was from writing to the memory containing the top of the while loop? |
| 01:51:04 | runvnc | that isn't inside of my try catch |
| 01:51:27 | Demos | right, but the segfault could be from you trying to write to ANYWHERE |
| 01:51:36 | runvnc | I just wish that somehow I could get an exception instead of segfault, but I can't, so that's that, and I will just have to be a little more careful and use a monitoring program to restart my process |
| 01:51:52 | Demos | not really, what you should do is not segfault |
| 01:52:05 | runvnc | as I said, I will just have to be more careful |
| 01:52:08 | Demos | right |
| 01:52:11 | runvnc | never segfaulting is a goal |
| 01:52:14 | runvnc | but for production |
| 01:52:20 | Demos | you should not segfault |
| 01:52:27 | runvnc | I must have a monitor for an occasional code path or situation that I did not think of |
| 01:52:47 | Demos | OK say we turned derefs of a null ptr into an exception, there is literally nothing useful you can do with that exception aside from crashing |
| 01:53:15 | runvnc | there could be a code path where that data was not initialized only in a special case |
| 01:53:29 | runvnc | whereas normal cases code continue |
| 01:53:58 | Demos | well then if that code path happens (and it is a null deref) than you get a stack trace and can eaisly fix the bug |
| 01:57:18 | runvnc | right but meanwhile I dropped 25 connections associated with child processes that had customers commands running on them |
| 01:57:43 | runvnc | I don't want to argue you with, and I'm not trying to say that it's nimrod's fault if I forget something basic like initializing data |
| 01:58:09 | runvnc | I just know that Nimrod seems to have advantages over most other environments, except in this case it does not have an advantage that other programming systems have |
| 01:58:48 | runvnc | which is that option to continue without completely exiting if there is an occasional code path with an unknown exception involving unintialized data |
| 02:00:37 | runvnc | well, not an exception, an unknown error I meant to say |
| 02:06:25 | Skrylar | that reminds me that there is something i would like to fiddle with at some point; putting a constraint solver on the VM |
| 02:10:00 | runvnc | maybe I could make a pragma that would wrap all of my if blah.len or whatever and add on a check for nil |
| 02:10:00 | * | zahary quit (Quit: Leaving.) |
| 02:13:22 | runvnc | maybe I can just use one of these templates http://forum.nimrod-lang.org/t/385 |
| 02:16:33 | EXetoC | it should be prevented obviously, and I think the language will be improved in this regard |
| 02:16:45 | * | q66 quit (Quit: Leaving) |
| 02:16:52 | EXetoC | there's "not nil" which might still be buggy/incomplete http://build.nimrod-lang.org/docs/manual.html#not-nil-annotation |
| 02:17:24 | runvnc | thanks |
| 02:18:00 | EXetoC | and then you have sequences/strings that must be initialized explicitly, but it seems pretty certain that that will change soon |
| 02:18:06 | EXetoC | did you at least get a stack trace? |
| 02:18:50 | runvnc | yup got a stack trace |
| 02:19:15 | runvnc | I am just lazy and used to node.js and coffeescript with ? and being very dummy-friendly |
| 02:27:23 | * | DAddYE joined #nimrod |
| 02:31:23 | Varriount | Meep |
| 02:31:33 | EXetoC | it's a nice shortcut |
| 02:31:34 | * | DAddYE quit (Ping timeout: 244 seconds) |
| 02:35:37 | * | Varriount quit (Ping timeout: 240 seconds) |
| 02:39:42 | runvnc | varriont: I am trying to use these suggestions http://forum.nimrod-lang.org/t/385 but the thing I am checking is a seq[string] and so it wont compile because that is written for a Node |
| 02:40:01 | runvnc | is there a way to change one of those templates so it will also work with a seq[string] |
| 02:40:26 | runvnc | my expression is if results?.len >0 |
| 02:41:49 | Demos | just initialize the seq |
| 02:42:19 | Demos | so instead of var foo: TSomeThingWithASeq you would say var foo = initSomethingWithASeq() |
| 02:42:39 | EXetoC | runvnc: he quit. use nick completion :-) |
| 02:43:40 | runvnc | demos its coming out of my module. so I can fix my module to return an empty sequence I guess |
| 02:43:59 | runvnc | I am just looking for an easy way to make these things safe without adding a lot of manual checks |
| 02:44:18 | runvnc | http://forum.nimrod-lang.org/t/385 this is perfect just doesn't work with sequences |
| 02:44:25 | runvnc | anyway I should do work instead of bothering you guys |
| 02:44:45 | Demos | you have it never be nill no matter what. unless it is actually optional in which case you usually want to do something interesting when it /is/ null |
| 02:50:43 | * | Varriount joined #nimrod |
| 02:51:00 | Varriount | runvnc: The thing is, nimrod doesn't run in a virtual machine like javascript, java, or python - it runs natively |
| 02:51:37 | EXetoC | well, there's that last example, which might not be feasible in conjunction with operators. either way, it doesn't seem to be handled very well by the compiler atm |
| 02:51:59 | Varriount | Which means that in most cases, when it accesses invalid memory, the OS stops the process and calls the signal handlers. |
| 02:52:38 | Varriount | (Although there are ways to recover from such cases, they tend to be tricky and very platform specific) |
| 02:52:46 | Skrylar | whats the conversation about? nillable types? |
| 02:54:27 | EXetoC | it seems to be more about shortcuts now |
| 02:57:13 | runvnc | http://forum.nimrod-lang.org/t/385/1#2134 |
| 02:58:53 | Varriount | runvnc: I'm glad you found my forum post useful though. |
| 02:59:10 | runvnc | yeah thanks |
| 03:00:07 | EXetoC | `@.` gives me an ICE with a familiar error message |
| 03:00:13 | Varriount | I looked at how C# programmers were oooh-ing and ahh-ing over the new safe access operator, and thought that there must be a way to do that for nimrod. |
| 03:00:33 | Varriount | Although, it won't work for ints, floats, or other base types. |
| 03:01:29 | runvnc | in c# they get an exception though right |
| 03:01:48 | runvnc | which they can have a wrapper for if they dont want their program to crash |
| 03:02:21 | runvnc | I mean without safe access |
| 03:02:46 | Demos | yes, but null pointer exceptions are /really/ dumb |
| 03:03:14 | Demos | *null pointer exceptions generated by a language runtime |
| 03:03:19 | EXetoC | I don't know why someone implemented it for D |
| 03:03:45 | EXetoC | someone must've had some use for it, but some did complain about it not making much sense |
| 03:03:48 | Demos | I think if you deref a null pointer in C while on windows with SEH on you will get an exception |
| 03:04:36 | Demos | in fact there are a lot of exceptions that should just never be caught to do anything but log and exit |
| 03:05:34 | runvnc | in my case the redis pop returned nil and if there was an exception in that while loop it could continue. wouldnt be something you would like, but would work prefectly |
| 03:05:39 | runvnc | perfectly |
| 03:05:52 | Demos | http://blogs.msdn.com/b/ericlippert/archive/2008/09/10/vexing-exceptions.aspx |
| 03:06:07 | Demos | that post just about sums up what I think about exceptions |
| 03:06:37 | Demos | also I think that if you disable exceptions (that is throw or raise is replaced with abort()) your program should still more or less work |
| 03:07:01 | EXetoC | runvnc: yes, if you'd handle it explicitly right there and then, in which you might as well just prevent it with some additional conditional statements |
| 03:08:26 | * | askatasuna quit (Ping timeout: 252 seconds) |
| 03:08:27 | runvnc | demos I have gone to some lengths to ensure that I can recover from restarts by using processing queues and multiple processes etc, however there are going to be cases where a core processes needs some state concerning multiple customers in memory in which case I would prefer that process did crash on the basis of a rare error I did not think of |
| 03:08:48 | runvnc | did not crash rather |
| 03:09:12 | runvnc | anyway I am going to take a shower |
| 03:09:30 | Demos | the thing is that you have a choice between crashing and working with possibly incorrect data |
| 03:10:07 | Varriount | runvnc: The error has to do with operator precedence I think |
| 03:10:08 | EXetoC | explosion of random failures (tm) |
| 03:10:25 | runvnc | those aren't the only two choices though, in many cases if I had some sort of exception, I could simply jump back to the top of my while loop and continue without any issue |
| 03:10:30 | * | askatasuna joined #nimrod |
| 03:10:47 | runvnc | ok thanks for the clue varriount |
| 03:11:21 | Varriount | runvnc: In any case, I don't think it will work with an if expression, because you can't do "nil > 0" |
| 03:11:43 | * | nande joined #nimrod |
| 03:12:02 | Varriount | That kind of operator is really more useful when doing attribute/member access. Sorry. |
| 03:15:26 | Varriount | runvnc: I'm gonna check what happens when I use Araq's new strongSpaces option on it. |
| 03:16:49 | Varriount | This is interesting. Apparently nimrod has a "braces" template filter.. I wonder what it does... |
| 03:18:06 | EXetoC | Varriount: braces? term-rewriting? |
| 03:18:33 | Demos | term rewriting macros sound like a good way to crash the compiler |
| 03:20:04 | flaviu | Is there any way to get a typeDesc of a type parameter? |
| 03:20:11 | EXetoC | at least some basic tests pass |
| 03:20:24 | Varriount | Demos: These are filters, not term rewriting macros. |
| 03:20:35 | EXetoC | flaviu: isn't it already a typedesc? |
| 03:20:48 | EXetoC | Varriount: where? |
| 03:21:03 | EXetoC | oh |
| 03:21:09 | EXetoC | the alternate syntax thing? |
| 03:21:19 | flaviu | I'm not sure, but my template requires a typeDesc to compile |
| 03:21:25 | Demos | flaviu, the semantics of typedesc are a little strange, and not written down anywhere |
| 03:21:46 | Varriount | EXetoC: https://github.com/Araq/Nimrod/commit/3ff9869e3d4fcca7802a673b09cb67797c62e2df#diff-b6862e8f17fa4c133ee3a5f3a2ba7cbbR23 |
| 03:21:49 | Demos | but type(T) /may/ do it |
| 03:21:55 | Demos | or actually just use typedesc[T] |
| 03:21:57 | EXetoC | flaviu: I don't know if your question makes sense then. perhaps you want tmpl(variable.type) |
| 03:23:22 | Varriount | Maybe filwit should implement his proposed syntax as a parser. |
| 03:23:22 | EXetoC | Varriount: right. I've expressed my reluctance towards that before. the last two anyway |
| 03:24:07 | Varriount | EXetoC: *shrug* |
| 03:25:55 | Varriount | runvnc: Your example works if I enable strongSpaces (although, if you add the () then if fails again. Probably something to discuss with Araq. |
| 03:26:45 | Varriount | "Works" being that the program is tripped up by the type mismatch for "nil > 0", rather than a syntax/template error. |
| 03:27:22 | EXetoC | I like Lua's "local x = y or z" feature |
| 03:27:46 | Varriount | EXetoC: What is that? |
| 03:29:54 | flaviu | Let me see if I can get a SSCCE for my template type parameter issue |
| 03:29:59 | EXetoC | Varriount: it assigns x the value of y if y != nil, or z otherwise |
| 03:30:11 | EXetoC | but that doesn't help in this case I think |
| 03:30:17 | Varriount | flaviu: SSCCE? |
| 03:31:12 | flaviu | A short example, "Short, Self Contained, Correct (Compilable), Example" |
| 03:33:55 | EXetoC | test case? :> |
| 03:34:09 | flaviu | Thanks, that sounds much better |
| 03:34:53 | * | brson quit (Quit: leaving) |
| 03:35:21 | * | brson joined #nimrod |
| 03:37:35 | * | askatasuna quit (Ping timeout: 252 seconds) |
| 03:38:47 | * | askatasuna joined #nimrod |
| 03:40:49 | flaviu | Ok, https://gist.github.com/flaviut/76225c559b0c80210df2 |
| 03:41:11 | flaviu | The top were there is a type parameter fails, but the bottom works |
| 03:41:44 | flaviu | Maybe I'm misunderstanding type inference, but I'd expect the top to work |
| 03:41:55 | Varriount | flaviu: I think that might be a problem with the VM/Generic handling |
| 03:42:50 | flaviu | Should I make a bug report? |
| 03:42:51 | Varriount | flaviu: It also might be because 'T' could be *anything* |
| 03:43:09 | Varriount | Including the meta types expr, statement, typedesc, etc |
| 03:43:52 | Varriount | flaviu: You might as well, at least people will then know about it. |
| 03:44:28 | EXetoC | [T...] for templates never worked for me, so I just assumed that it doesn't work |
| 03:47:46 | flaviu | EXetoC: Shouldn't that be reported? IMO, it should be documented if it isn't supposed to work |
| 03:52:56 | runvnc | --nilChecks:on ?? |
| 03:53:58 | Varriount | runvnc: That's not a compiler option... is it? |
| 03:54:19 | runvnc | I dunno this guy said it is http://forum.nimrod-lang.org/t/165/2 |
| 03:54:35 | runvnc | it is listed here http://nimrod-lang.org/manual.html#compilation-option-pragmas |
| 03:55:41 | Varriount | runvnc: Also note "Note: This document is a draft! Several of Nimrod's features need more precise wording. This manual will evolve into a proper specification some day." |
| 03:56:36 | Varriount | But... looking at the compiler code, there does appear to be something about nil checks |
| 03:56:54 | runvnc | lol |
| 03:57:15 | runvnc | well it says invalid compiler option |
| 03:58:02 | Varriount | I think it's a pragma.. |
| 03:58:36 | runvnc | oh maybe that means the thing exetoc mentioned like not nil |
| 03:58:59 | runvnc | although why would you need to generate code for that since it was doing it at runtime |
| 03:59:55 | Varriount | Theres also "objCheck" |
| 04:00:22 | EXetoC | runvnc: that runtime code needs to be generated then |
| 04:00:28 | * | flaviu quit (Remote host closed the connection) |
| 04:01:00 | runvnc | let me try objcheck |
| 04:01:32 | Demos | not nil is intended to be a compile time thing (I hope) |
| 04:02:28 | Varriount | Yeah. Sorta |
| 04:02:57 | Varriount | I'm looking through pieces of code. A lot of the occurences of "nilcheck" occur in the C generation files. |
| 04:03:21 | Demos | if it is a runtime check than I would much rather it omit the check and I just get a segfault |
| 04:03:23 | Varriount | Maybe Nimrod once had a feature to turn off isNil() and friends? |
| 04:03:46 | Varriount | Demos: It's equivalent to checking if a pointer is null |
| 04:04:30 | Demos | that is not what I want from it, I want a compile time garentee that is it not nil |
| 04:04:36 | EXetoC | I don't know how you'd do that |
| 04:04:48 | Demos | actually you could just use a var |
| 04:05:01 | Demos | come to thing of it :D |
| 04:05:08 | EXetoC | (varriount) |
| 04:05:21 | EXetoC | Demos: so "not nil" |
| 04:05:25 | Demos | although var has a really annoying tendency to go away really fast since we don't have var vars |
| 04:05:29 | Demos | nope |
| 04:05:57 | Varriount | Demos: Stop being paranoid about performance, or I will feed you to #python |
| 04:06:08 | runvnc | varriount so the objChecks and nilChecks don't have code tied to them now? |
| 04:06:28 | Varriount | runvnc: They do.. but it looks incomplete. I'm skimming through things right now. |
| 04:06:33 | runvnc | objChecks is accepted as a parameter to nimrod but doesn't check for nil when I access this thing |
| 04:06:58 | Demos | we are a systems programming language, and you are right not nil should be a runtime check, but I am not sure when you would use not nil over a var param |
| 04:07:45 | Varriount | What does objChecks do then...? |
| 04:08:14 | runvnc | you have every other feature I want to use it as an applications programming language, whatever the distinction is, I believe that nilChecks could even be considered valid in a systems context |
| 04:08:25 | runvnc | but certainly is valid in the context of a modern application programming system |
| 04:08:45 | runvnc | I could try objChecks with a different type of reference or something |
| 04:09:25 | runvnc | probably checks objects, I am using a seq[string] |
| 04:09:37 | Demos | I am all for a compiler option that turns on null checks |
| 04:09:47 | runvnc | there is not nil already |
| 04:09:52 | runvnc | 'not nil |
| 04:09:54 | runvnc | ' |
| 04:10:11 | Demos | I think I was wrong in objecting to a null check on not nil params |
| 04:10:22 | Demos | but I still would rather use a var param |
| 04:11:11 | Varriount | runvnc: It looks as if the 'nilChecks' option isn't hooked up to the command line parsing code. |
| 04:11:30 | runvnc | maybe he's not done |
| 04:11:35 | runvnc | or morally against it like demos |
| 04:11:59 | runvnc | thanks for looking into that for me |
| 04:13:28 | Varriount | runvnc: No problem. I'm actually looking to see if I can enable to the code. |
| 04:14:05 | Varriount | if optNilCheck in p.options: |
| 04:14:05 | Varriount | linefmt(p, cpsStmts, "#chckNil((void*)$1);$n", addrLoc(loc)) |
| 04:14:23 | EXetoC | might as well just test the actual pragma first |
| 04:15:18 | EXetoC | Demos: I don't think it has the same guarantees |
| 04:15:47 | * | carum joined #nimrod |
| 04:16:51 | Demos | EXetoC, a var param should never be nil |
| 04:20:35 | * | carum quit (Ping timeout: 244 seconds) |
| 04:21:04 | runvnc | why did kt write this http://forum.nimrod-lang.org/t/165/2 was that enabled when he wrote it |
| 04:23:31 | Varriount | Lets see.. |
| 04:24:58 | * | Varriount turns his time machine on |
| 04:25:35 | Varriount | runvnc: How many months ago was that post? |
| 04:27:49 | runvnc | right. many |
| 04:28:10 | Varriount | :P |
| 04:28:30 | runvnc | can I put that linefmt somewhere in the nimrod compiler code then and get that |
| 04:28:31 | Varriount | runvnc: I've checked out a copy of the code at the time that post was made. |
| 04:28:57 | runvnc | awesome |
| 04:28:59 | Varriount | runvnc: It's already in the compiler code. And I wouldn't advise it. |
| 04:29:46 | runvnc | well I mean I appreciate your looking into this, I hope you are doing it for sake of experimentation and not just on account of me complaining so much |
| 04:29:53 | runvnc | I have to go get dinner |
| 04:31:48 | Varriount | runvnc: Experimentation. |
| 04:32:11 | Varriount | <3 'koch temp' |
| 04:33:51 | * | Demos_ joined #nimrod |
| 04:33:58 | * | askatasuna quit (Ping timeout: 240 seconds) |
| 04:39:22 | Varriount | Hi again Demos |
| 04:47:03 | Demos_ | hi, I think I am logged on from both my computers |
| 04:47:06 | Demos_ | fowl: ping |
| 04:47:37 | Demos_ | yup that ping kicked my netflix out of fullscreen mode |
| 04:48:29 | * | brson quit (Ping timeout: 240 seconds) |
| 04:54:09 | * | askatasuna joined #nimrod |
| 04:56:50 | * | askatasuna quit (Client Quit) |
| 04:58:30 | * | [1]Endy joined #nimrod |
| 05:15:59 | Skrylar | Meh. I wish we could fire people who make FPS games so they go make better games in things. |
| 05:19:08 | * | xenagi joined #nimrod |
| 05:22:06 | Demos | I like FPS games |
| 05:22:58 | Skrylar | I tend to just ignore them almost completely; or really anything multiplayer anymore. |
| 05:23:31 | Demos | payday 2 is fun |
| 05:23:45 | Skrylar | Dealing with pubbies sucks ass and I tried joining some groups, they weren't much better. |
| 05:23:48 | Demos | me and my friends have been playing lots of phantasy star online 2 recently as well... |
| 05:23:59 | Skrylar | Way more concern for making forum posts on brag sites than actually doing anything |
| 05:24:12 | Skrylar | bleh.. MMOs |
| 05:24:19 | Demos | I also sometimes play ArmA with United Operations, they are a public group that is happy to play with anyone who actually wants to work as a team |
| 05:24:44 | Skrylar | I remember someone I used to know was plugging Arma groups pretty heavily |
| 05:24:57 | Skrylar | I poked him about it and it was just "oh well go fill out these forms" and it was like "yeah.. no." |
| 05:25:19 | Demos | UO is pretty much like "register on our forums, connect to TS, play" |
| 05:25:30 | Skrylar | Its been my experience the communities you get personally invited to because someone played near you tend to be the ones you have fun in, and the ones that you have to fill application forms for tend to care more about pomp |
| 05:25:33 | Skrylar | but thats just my experience |
| 05:26:09 | Demos | UO is very good, as long as you are willing to try and work as a team and learn from your mistakes |
| 05:26:21 | Skrylar | i do that, but thats never been my problem |
| 05:26:34 | Demos | they have a training event tomorrow night (EST), maybe you could give them a shot |
| 05:26:52 | Skrylar | I just physiologically don't have the reflexes to play FPS games effectively, which means I get jammed as the medic or guy who stands around repairing shit |
| 05:27:04 | Demos | dude, this is ArmA |
| 05:27:06 | reactormonk | Skrylar, perfect. |
| 05:27:38 | Demos | we are talking about a game that has realistic simulation of radios, with distortion, terrain blockage and everything |
| 05:27:45 | Skrylar | i should reinstall DE3 later, that was a fun experience |
| 05:27:55 | Demos | DE3? |
| 05:27:59 | Skrylar | deus ex |
| 05:28:09 | Skrylar | the original and human revolution were fun |
| 05:31:11 | Skrylar | Kinda sad that the TA-style RTS games aren't more popular than they are; they don't rely as heavily on raw reflex as most of the RTS genre has decided it needs to |
| 05:33:26 | Demos | well PA is a thing, I dont really like it that much though |
| 05:34:09 | Skrylar | I bought that but... ergh. |
| 05:34:23 | Skrylar | The GUI is terrible |
| 05:34:57 | Demos | it is just a big transparent webpage |
| 05:35:11 | Skrylar | I feel like RTS keeps thinking about how to make things a bigger pain, because there are a lot of them deciding they need to be DOTA clones but they don't improve the actual means of command |
| 05:35:49 | Skrylar | and there isn't a pressure to improve anything in the games industry; people just say "oh its competitive" when you criticize bad design |
| 05:36:32 | renesac | Skrylar, cool to see siphash working |
| 05:36:50 | Skrylar | renesac: yerp, i need to make a github for that later |
| 05:36:58 | renesac | did you benchmark against the C version? |
| 05:37:05 | Skrylar | no |
| 05:37:08 | Skrylar | do we have a benchmarking unit |
| 05:37:18 | EXetoC | RTS dota clone? sounds about right |
| 05:37:20 | renesac | ? |
| 05:37:28 | renesac | I'm making a benchmark module |
| 05:37:33 | Skrylar | i ported over the test vectors to make sure the siphash works correctly, but there isn't a benchmark |
| 05:37:34 | renesac | but still very incomplet |
| 05:37:55 | renesac | https://github.com/ReneSac/Benchmark |
| 05:38:07 | renesac | and no calling of external processes either |
| 05:38:16 | Skrylar | EXetoC: yeah there's been a few try that; Tryst with the 'two unit modes' and Company of Heroes 2 decided to go further in to the MOBA control style of micro-over-macro |
| 05:38:36 | renesac | though you can call it yourself of course |
| 05:38:48 | Skrylar | Nobody is interested in improving macro, which is kind of funny because you would expect PA to be a macro-oriented game yet it still controls like a crappy starcraft |
| 05:38:59 | renesac | and bench around that |
| 05:39:14 | renesac | I still haven't finished the 'timeit' lookalike template either... |
| 05:41:33 | Demos | yeah, there are grand strat games but they are a little ... too macro |
| 05:41:59 | Skrylar | Those are still micro games :/ |
| 05:42:13 | Demos | I guess macro is harder to do than micro |
| 05:42:23 | Skrylar | The funny thing is when you see someone say they don't want micromanagement, and someone says Civ, but Civ is entirely micromanagement |
| 05:42:42 | Skrylar | spreadsheet optimizing cities across the map every turn, etc |
| 05:43:36 | Skrylar | I think Hearts of Iron 3 tried to lessen that by having it so once units were put in a division they would run with some degree of autonomy within that regiment |
| 05:44:43 | Skrylar | just becomes funny when you end up looking at the concept of spending 20 hours memorizing math so you can play a game and enjoy it or you could do something productive . . . lol |
| 05:47:06 | NimBot | nimrod-code/nimbuild master a4aa4fa Andreas Rumpf [+0 ±1 -0]: make it compile with devel |
| 05:47:06 | NimBot | nimrod-code/nimbuild master cdb3b7a Andreas Rumpf [+0 ±1 -0]: Merge branch 'master' of https://github.com/nimrod-code/nimbuild |
| 05:56:24 | Demos_ | so I am seeing {.push: cdecl, dynlib: name.}, it is not compileing. Does the {.push.} pgrama even work for stuff that is not compiler options? |
| 05:59:16 | Araq | get rid of the ':' after the 'push' |
| 05:59:49 | Demos_ | but it does what I expect? |
| 06:00:06 | Araq | push only affects proc declarations |
| 06:00:14 | Demos_ | OK that is fine with me |
| 06:00:19 | Araq | so a proc type won't get the 'cdecl' |
| 06:00:25 | Araq | watch out |
| 06:01:25 | Araq | but yes, it does work, we have wrappers that use it |
| 06:01:48 | * | r0b1 quit (Quit: brb) |
| 06:05:16 | fowl | Demos, png |
| 06:05:28 | EXetoC | gif |
| 06:05:41 | fowl | Demos_, you have to push callconv: cdecl |
| 06:06:12 | Araq | really? I don't think so |
| 06:06:29 | Araq | but push cdecl is limited in the way I wrote |
| 06:07:30 | fowl | yes |
| 06:08:29 | Araq | aporia.nim(699, 12) Error: undeclared identifier: 'set_upper' |
| 06:08:42 | Araq | babel update gtk2 |
| 06:08:44 | Araq | Downloading package list from gtk2 |
| 06:08:45 | Araq | Error: unhandled exception: Der angeforderte Name ist g├╝ltig, es wurden jedoch keine Daten des angeforderten Typs gefunden. |
| 06:08:47 | Araq | -.- |
| 06:09:13 | Demos_ | anyway, fowl I just fixed your assimp wrapper |
| 06:09:16 | Demos_ | and submitted a pr |
| 06:09:42 | Skrylar | lol ass imp |
| 06:09:49 | * | brson joined #nimrod |
| 06:09:59 | Skrylar | that library name amuses me every time i see it |
| 06:10:08 | Demos_ | also, your wrappers have no license |
| 06:11:10 | fowl | can you move it to a new repo |
| 06:11:18 | Araq | babel list |
| 06:11:19 | Araq | Error: unhandled exception:.babel\packages.json(1, 0) Error: { expected [EJsonParsingError] |
| 06:11:26 | Demos_ | fowl: the assimp wrapper? |
| 06:11:31 | fowl | yea |
| 06:11:35 | Araq | hmm babel doesn't like me anymore |
| 06:11:45 | Demos_ | I was going to but I wanted your permission first. |
| 06:13:23 | Araq | hmm my packages.json is empty ... |
| 06:13:49 | fowl | Demos_, goo for it |
| 06:14:45 | Demos_ | OK, mind if I stick a BSD license on it? |
| 06:15:09 | Araq | Demos_: default here is MIT, just sayin |
| 06:15:21 | Demos_ | allright I can do MIT |
| 06:15:35 | EXetoC | Araq: have you done "babel update"? it's not done automatically IIRC |
| 06:15:46 | Demos_ | for wrappers I usually use the same lecense as the library I am wrapping |
| 06:15:52 | Araq | EXetoC: I just did |
| 06:16:01 | Araq | Demos_: that makes perfect sense too |
| 06:16:04 | EXetoC | :E |
| 06:16:11 | fowl | Demos_, bsd is cool |
| 06:16:27 | Araq | how do I update gtk2 via babel? |
| 06:16:33 | Demos_ | that was mostally for when I wrapped libalpm which was GPLv3 |
| 06:16:38 | * | xenagi quit (Remote host closed the connection) |
| 06:16:38 | Araq | "babel update" is only for the packages list |
| 06:16:47 | fowl | babel install gtk2 |
| 06:17:32 | Araq | gtk2 already exists. Overwrite? [y/N] |
| 06:18:14 | Araq | but ok, this works |
| 06:18:59 | Araq | it's also as annoying as the other package managers I'm used to, so we're doing a good job :-( |
| 06:20:24 | Araq | I need to talk with dom96 about babel's command line interface, babel update should update all the packages that I have installed |
| 06:20:26 | fowl | i will be confused the day you find some software that you like |
| 06:20:27 | EXetoC | use --accept/--reject to handle that automatically |
| 06:21:08 | EXetoC | yeah there are many conveniences that are yet to be added |
| 06:21:53 | Araq | which reminds me ... I wanted to implement some kind of natural language processing for command line interfaces |
| 06:21:55 | Demos_ | I have been meaning to try and see how one would add nimrod libraries to the aur |
| 06:22:23 | Demos_ | Araq: that would be neat if integrated into zsh like completion functionality |
| 06:23:49 | Varriount | Ohi Araq |
| 06:24:30 | Varriount | Araq: What does the 'bracket' standard template filter do? |
| 06:24:35 | Varriount | *brackets |
| 06:26:04 | runvnc | Natural language processing for command line interfaces.. seriously? Sounds awesome but not sure if serious heh |
| 06:29:18 | runvnc | http://en.wikipedia.org/wiki/Attempto_Controlled_English |
| 06:30:08 | Skrylar | if you use pigeon english then yeah you can parse it, but thats not what people want |
| 06:30:30 | Skrylar | if you claim to support english but its actually a specific subvariant it just confuses people because its too close to what they would have otherwise written |
| 06:30:43 | Varriount | I know of a python module called pyopt that turns a command line description into a parser. |
| 06:30:46 | runvnc | that ace thing looks pretty powerful. pretty big download and set of related tools though |
| 06:31:10 | Skrylar | the Inform project has a page on how their NLP system works which is interesting, but eh |
| 06:31:26 | EXetoC | Varriount: docopt? |
| 06:31:37 | Varriount | EXetoC: Er yes, that. |
| 06:31:37 | EXetoC | ok I found another pyopt |
| 06:31:46 | Skrylar | Inform is a weird creature; its like a simplified NLP parser with some prolog-like logic trees under it |
| 06:32:33 | Varriount | EXetoC: A couple of weeks ago I was looking for docopt, found pyopt, then found docopt, and then got the two confused just now. |
| 06:33:41 | Varriount | Araq: Your strongspaces feature came in handy today. runvnc and I were looking at a safe access operator template, and strongspaces fixed things. |
| 06:37:39 | Demos_ | hm it looks like packages.json may have a trailing ',' when it should not have |
| 06:37:43 | Demos_ | does not seem to be a problem though |
| 06:42:11 | * | [1]Endy quit (Ping timeout: 264 seconds) |
| 06:43:18 | Demos_ | OK assimp is its own repo and I added it to my fork of packages and submitted a pr for the main repo |
| 06:43:40 | runvnc | maybe TextBlog could be converted to c with cython or something |
| 06:45:45 | Varriount | Demos: Does that package import textures as well as models? |
| 06:46:01 | Demos_ | no. assimp may but my wrapper does not |
| 06:46:08 | Varriount | Aw.. |
| 06:46:18 | Demos_ | there is a wrapper for libpng in babel though! |
| 06:47:10 | Demos_ | actually it looks like it does |
| 06:47:11 | Demos_ | sorta |
| 06:47:21 | Varriount | Demos: I'm thinking about porting parts of the libGDX api as I go through the various opengl tutorials. Some of that includes texture loading from images. |
| 06:48:28 | Demos_ | I have been thinking of using freeimage |
| 06:48:42 | Demos_ | another option is to write a TGA loader, targa is a very simple format |
| 06:49:15 | Demos_ | you can essentially just copy the spec |
| 06:50:40 | Demos_ | OK assimp only imports textures if they are stored as part of the file format that it is importing |
| 06:51:03 | Varriount | :< |
| 06:51:27 | Demos_ | anoyne have thoughts on a good multi-format image loading library, I hear freeimage is decent |
| 06:52:11 | Varriount | Demos: What about SOIL? |
| 06:53:56 | Demos_ | freeimage sounds more "industrial strength" SOIL is probably going to be eaiser to wrap though |
| 06:55:29 | Varriount | At least, for opengl purposes. |
| 06:55:51 | Varriount | Hm. But will it work on android? |
| 06:56:32 | EXetoC | I tried to wrap freeimage, but the source code is just awful |
| 06:59:21 | fowl | Varriount, use devil for importing textures |
| 06:59:40 | Varriount | fowl: Isn't that GPL though? |
| 06:59:57 | fowl | does that mean you cant use it? |
| 07:00:15 | EXetoC | LGPL |
| 07:00:27 | Varriount | Well, what if I want to sell an amazing popular 2d platformer someday? :3 |
| 07:00:50 | Demos_ | I thought DevIl was unmaintained these days |
| 07:00:58 | Demos_ | and freeimage headers look decent to me |
| 07:02:22 | Varriount | I prefereably need something that doesn't use any deprecated opengl features (Below 3.0) |
| 07:03:33 | Demos_ | image loading should not use /any/ openGL features |
| 07:04:59 | Araq | Varriount: can you please make gist about the non-strongSpaces issue? |
| 07:05:18 | Demos_ | anyway I am going to go to sleep. Maybe I will wrap freeimage tomorrow |
| 07:06:03 | Skrylar | Demos_: not exactly. devIL and SOIL are meant for loading images directly to GL |
| 07:06:19 | Skrylar | the whole point of those two libs is that you just ask it to put the file in a gltexture |
| 07:06:29 | fowl | Demos_, theres a fork of devil but i dont see it on AUR, so it might not actually exist |
| 07:06:31 | Varriount | Araq: Erm, You could just check the latest forum post -> http://forum.nimrod-lang.org/t/385 |
| 07:07:09 | Varriount | You would think image loading libraries would be more plentiful |
| 07:07:11 | Araq | Varriount: ah I see |
| 07:07:38 | Varriount | Araq: Although, it doesn't work if you add the () to len - could you explain why? |
| 07:08:23 | Araq | because it's about '?' vs' >' ? |
| 07:09:05 | Araq | ? has lower precedence than > in the table in the manual |
| 07:09:30 | fowl | Varriount, everybody uses devil |
| 07:10:32 | Varriount | Araq: Yeah, but what about the function call? |
| 07:10:52 | Demos | fowl, looks like there is something called "ResIL" that is being maintained |
| 07:11:21 | fowl | Demos, yea but its not on AUR so my motivation for wrapping it is very flaccid |
| 07:11:50 | Demos | oh god devil has an openGL interface, why the hell would you do that |
| 07:11:57 | Demos | or rather opengl style |
| 07:12:29 | * | Demos_ quit (Ping timeout: 240 seconds) |
| 07:13:24 | Araq | Varriount: wtf what happened with --nilChecks ? |
| 07:13:34 | Araq | you said you found the offending commit? |
| 07:13:46 | Varriount | No, I didn't. |
| 07:13:52 | runvnc | Maybe I didn't know how to test it |
| 07:14:02 | runvnc | seemed to say wasn't a valid flag though |
| 07:14:06 | runvnc | on my compiler |
| 07:14:11 | Varriount | And on mine. |
| 07:15:10 | Demos | biect that mofo |
| 07:15:13 | Demos | *bisect |
| 07:15:18 | Demos | I really need to sleep |
| 07:15:21 | * | Demos quit (Quit: Leaving) |
| 07:15:25 | fowl | meh |
| 07:15:33 | fowl | dont you want to call it assimp instead of nim-assimp?? |
| 07:15:38 | fowl | oh he left |
| 07:15:54 | Araq | runvnc: nope, it fails for me too |
| 07:16:09 | Araq | and even stranger, it's not documented either |
| 07:16:50 | runvnc | I don't know if that applies to my thing exactly, but if that flag could result in fewer segfaults or whatever when I screw up or the possibility of getting an exception instead, would be nice for me in some cases |
| 07:17:36 | Varriount | runvnc: nilValueException? :3 |
| 07:17:51 | runvnc | heh I dunno |
| 07:18:07 | * | carum joined #nimrod |
| 07:18:11 | runvnc | whatever the flag is supposed to do |
| 07:19:28 | Araq | it turns on nil checks in some places |
| 07:19:45 | Araq | but very rarely so it won't help you |
| 07:20:03 | Araq | maybe that's why the former araq removed it |
| 07:20:36 | fowl | we have a new araq? |
| 07:21:11 | Araq | yes. An araq with even less sleep |
| 07:21:23 | runvnc | oh so it just turned out that there were a lot of places to nil check and to have that flag would be confusing/broken unless they were all implemented |
| 07:21:57 | runvnc | maybe not feasible to actually implement |
| 07:22:06 | EXetoC | fowl: I don't like the idea of using the exact same name as the original |
| 07:22:31 | fowl | runvnc, nil checking should be left to the user |
| 07:22:50 | fowl | segfaults arent that bad when you get backtraces |
| 07:23:09 | * | carum quit (Ping timeout: 252 seconds) |
| 07:23:14 | Araq | runvnc: it's pretty easy to implement in the compiler, but since it was incomplete I removed it, fearing the bug reports |
| 07:23:15 | fowl | EXetoC, why not? its an interface to that library |
| 07:23:17 | Skrylar | I like the idea of nillable/non-nillable types :\ |
| 07:23:38 | Araq | Skrylar: we got these but it's a big update on the stdlib |
| 07:23:53 | Skrylar | i was disagreeing with fowl mostly, Araq |
| 07:24:01 | Araq | and you're talking about compile-time checks |
| 07:24:12 | Araq | --nilChecks were runtime checks |
| 07:24:16 | runvnc | I have to give my honest opinion, being someone who has avoided c for many years (since I was in high-school/college) and having gotten used to being able to keep my server running even if I make a mistake that appears on an occassional code path |
| 07:24:45 | runvnc | almost everything about Nimrod from an application programmer standpoint seems wonderful, but the segfaulting is unfriendly and unnecessary |
| 07:25:00 | Varriount | runvnc: What have you been using prior to nimrod? (Please don't say PHP) |
| 07:25:09 | Skrylar | node.js i thought |
| 07:25:10 | runvnc | so maybe I will be the only one but I would certainly appreciate this nilcheck thing |
| 07:25:22 | Skrylar | i think the last time i segfaulted was when i didn't initialize a sequence |
| 07:25:40 | Araq | Skrylar: that will be fixed in a different way |
| 07:25:48 | Skrylar | which i will admit is un-awesome, but there's not a whole lot that the compiler could have done to help with it |
| 07:25:50 | runvnc | I most recently programmed in Node.js, with ToffeeScript, previously CoffeeScript, previously JavaScript, previously C#/.NET, previously PHP, previously C++ |
| 07:25:52 | fowl | doesnt runtime nil checking require a check every time a pointer is dereferenced? |
| 07:26:09 | runvnc | yes I am getting a segfault on an seq of strings that wasn't inited |
| 07:26:11 | Skrylar | well what they do in rust is use maybe's |
| 07:26:22 | Araq | len for a nil seq will return 0 and 'add' will work with nil |
| 07:26:28 | Skrylar | if something is allowed to be null you have to case x of some(y) => foo, none() |
| 07:26:33 | fowl | my maybe() works for that |
| 07:26:45 | runvnc | ok well great so the particular problem I actually had will go away if you make it work that way with seq |
| 07:27:02 | Araq | Skrylar: imho that's way too annoying so nimrod does it differently |
| 07:27:02 | runvnc | awesome |
| 07:27:08 | fowl | when compiles(val.isNil): result = (val.isNil, val) |
| 07:27:11 | Varriount | Add will work for nil...? |
| 07:27:26 | Skrylar | Araq: doesn't the current nil-tracker require if x != nil etc |
| 07:27:27 | fowl | not val.isnil * |
| 07:27:40 | Araq | Skrylar: yes but it's slightly smarter |
| 07:28:01 | Skrylar | Araq: i still think it would be neat to mangle the VM so it can run like klee does |
| 07:28:21 | Araq | please elaborate |
| 07:28:47 | Skrylar | Klee is a symbolic runtime for LLVM; you give it a harness, mark symbolic parameters, and it steps through the function call with an STP solver to break it |
| 07:29:02 | Skrylar | e.g. it noses around for "if x > y" and then tries to find edge cases that weren't checked |
| 07:29:29 | Skrylar | all though klee does a full-depth check which makes it unsuitable for large projects; i think it would be nicer if it could be confined to one specific function and mock sub-calls |
| 07:29:54 | Skrylar | but basically, it uses a constraint solver and tries to find parameters that break your function |
| 07:30:11 | Skrylar | then produces those values so you can write a unit test on it |
| 07:30:46 | Araq | that's cool but we won't get it anytime soon |
| 07:31:07 | Araq | unless you use Klee on the produced C code ;-) |
| 07:31:33 | Skrylar | i don't tend to use klee; its neat but it wants to scan the whole program which is kind of silly |
| 07:31:57 | Araq | dunno, whole program compilation is the future |
| 07:32:13 | Araq | and is what nimrod does, but we need to change that |
| 07:32:22 | Skrylar | its not a whole program compilation issue |
| 07:32:29 | Skrylar | its an issue that you can't confine it to test a specific module |
| 07:32:39 | Skrylar | it tries to do branch checking on the whole program from the start |
| 07:33:01 | Skrylar | you can't tell it "I have 100% test coverage on that module, please don't make 500 sub-branches to test it again" |
| 07:33:11 | Varriount | Araq: Could you explain the reasoning behind changing the behavior of uninitialized sequences? |
| 07:34:09 | Araq | Varriount: lazy initialization of seqs and strings can be very important for performance and the 'nil' checks in len and 'add' are *cheap*, len should be able to use a conditional move |
| 07:34:40 | Varriount | Araq: What will adding to a uninitialized sequence do? |
| 07:34:55 | Araq | treat nil as @[] |
| 07:35:08 | Varriount | So.. it will initialized the sequence? |
| 07:35:13 | fowl | what about nilseq[0] |
| 07:35:13 | Araq | yes |
| 07:35:33 | Varriount | Araq: Oh good. I thought maybe you planned to just have add() be a no-op |
| 07:36:01 | Araq | Varriount: you should know me better by now :P |
| 07:36:17 | Varriount | True. You're not a PHP programmer. |
| 07:36:34 | fowl | lol |
| 07:36:54 | fowl | i think @ mightfail() should be supported |
| 07:37:03 | Araq | There is nothing worse than a program that doesn't crash but instead produces wrong results. |
| 07:38:00 | Varriount | Araq: Behold, sheer insanity - https://eval.in/111886 |
| 07:38:52 | Skrylar | Varriount: seems legit |
| 07:38:59 | Skrylar | QNKCDZO is obviously the NSA key |
| 07:39:12 | Araq | Skrylar: lol |
| 07:39:34 | Araq | fowl: what's @mightfail() ? |
| 07:39:44 | fowl | Araq, it swallows errors |
| 07:40:07 | Varriount | Also, evidence that prolonged exposure to PHP does terrible, terrible things to the human brain -> http://forums.devshed.com/php-development-5/php-fractal-bad-design-hardly-929746.html |
| 07:43:19 | EXetoC | Araq: yeah, what about nilseq[0]? |
| 07:44:08 | fowl | or seq.items |
| 07:47:04 | EXetoC | implicit initialization was discussed before |
| 07:47:45 | Araq | nilseq[0] continues to crash |
| 07:48:04 | Araq | what about it? its length is 0 |
| 07:48:19 | Araq | ok, might also raise an EInvalidIndex exception |
| 07:48:55 | Araq | but these are second class exceptions btw as they are gonne in release mode and the spec doesn't require them to be raised |
| 07:49:12 | Araq | the spec allows a fatal error instead too |
| 07:49:25 | Araq | er ... well I need to write that part of the spec |
| 07:50:52 | Varriount | Goodnight guys. May all your syntactic ambiguities be resolved. |
| 07:53:49 | EXetoC | ok well if length works like that |
| 07:53:51 | Skrylar | ugh. i forgot there was another problem with his PHP code potentially |
| 07:54:06 | Skrylar | there's a thing where you never use a comparer that exits early in security code |
| 07:55:25 | Skrylar | i don't have the link on hand but i recall an article talking about timing attacks which made use of cmp() exiting early, so they could make educated guesses about how their brute forces were succeeding |
| 07:59:02 | * | flyx read the nimrod tutorial. most things look nice, but generics look like C++ templates |
| 07:59:34 | flyx | is it possible to constrain generic parameters? |
| 08:02:51 | EXetoC | they will share some aspects obviously |
| 08:02:54 | EXetoC | flyx: http://build.nimrod-lang.org/docs/manual.html#type-classes |
| 08:03:31 | EXetoC | you should see "T: int|string" there |
| 08:04:17 | flyx | ah, nice |
| 08:07:10 | EXetoC | scroll down and you'll see "User defined type classes" which allows for more detailed constraints among other things. it's new and relatively untested though |
| 08:08:28 | flyx | wow |
| 08:08:49 | flyx | there are quite some features in nimrod I've never seen before |
| 08:09:34 | flyx | well, mostly the compile-time things |
| 08:11:23 | * | DAddYE joined #nimrod |
| 08:11:56 | flyx | is there any preference of a build system? plugins for existing build systems? |
| 08:17:32 | EXetoC | flyx: I only know of this https://github.com/fowlmouth/nake |
| 08:17:54 | EXetoC | caching is done by the compiler |
| 08:24:10 | flyx | I'm sceptical about nimrod's approach to compilation, or I may not fully understand it yet |
| 08:24:25 | flyx | so I only give one file and the compiler automatically discovers all other files? |
| 08:25:09 | flyx | what if I need different flags for different files? |
| 08:26:27 | Araq | you add .push and .pop pragmas to the files that need it then |
| 08:28:08 | * | brson quit (Quit: leaving) |
| 08:28:35 | flyx | I meant more like, defines used in when statements, that are not known in the file |
| 08:30:17 | Araq | not sure what you mean |
| 08:32:47 | Araq | Varriount: to answer your questions: braces and endx parsers have never been implemented |
| 08:33:30 | flyx | well I just started reading the docs, so I may not fully understand about how everything works |
| 08:34:08 | flyx | I assumed values used in "when" statements can be given on the command line |
| 08:34:35 | flyx | so I wondered, how could I set such a value for only one source file if everything is compiled at once |
| 08:35:05 | Araq | you can use --define:foo and when defined(foo) and you're right that it is a global switch then |
| 08:35:25 | Araq | but it's not a problem in practice afaict |
| 08:35:45 | Araq | 'defined' has other problems |
| 08:35:49 | flyx | okay. I might be too familiar with "classical" one-file-at-a-time-compilation |
| 08:37:26 | flyx | what about headers? there don't seem to be any in nimrod, how can I use a nimrod-compiled library in my code? |
| 08:38:23 | Araq | read the tutorial or the manual then, keywords: module, import, export marker * |
| 08:43:08 | flyx | I saw that, but what happens when I import something which is not available as source, but only as compiled library? does the compiler "extract" the exported definitions from the library and make them available in my code? |
| 08:45:17 | EXetoC | there's --app:lib and --app:staticlib |
| 08:45:55 | flyx | sure. I understand how I can create a library, but I don't really understand how to use it |
| 08:45:57 | EXetoC | which doesn't answer the question :p |
| 08:46:38 | flyx | let's say I want to spread a nimrod library as binary package. I include the nimrod-compiled binary and nothing else? |
| 08:48:11 | flyx | the developer using the library also needs a way of knowing the exports of the library, so I reckon I have to include *some* form of include/header files |
| 08:49:18 | Araq | flyx: there are symbol files which you can distribute instead of nimrod files, except that they are currently broken |
| 08:49:30 | flyx | ah |
| 08:49:42 | Araq | they are also quite easy to disassemble but we can fix that |
| 08:50:56 | flyx | is nimrod on the whole a spare-time/volunteer project? |
| 08:51:02 | EXetoC | most people just import complete modules though |
| 08:51:19 | EXetoC | including the standard library |
| 08:51:46 | flyx | well that works for small projects |
| 08:52:51 | EXetoC | I think it works for most projects, assuming that the source code is indeed available |
| 08:53:35 | flyx | I mean, look at C++, the folks there are doing stuff like precompiled headers to speed up compilation, because it takes a fair amount of time for bigger projects. |
| 08:55:25 | EXetoC | this will obviously be dealt with at some point, but keep in mind that nimrod code compiles a lot faster |
| 08:56:05 | * | _fel1x quit (*.net *.split) |
| 08:56:05 | * | oxful quit (*.net *.split) |
| 08:57:35 | flyx | well. so far, nimrod looks promising |
| 08:57:41 | flyx | I might want to give it a try |
| 09:03:05 | * | _fel1x joined #nimrod |
| 09:03:26 | * | oxful joined #nimrod |
| 09:10:23 | EXetoC | flyx: ok. feel free to put that limitation on the record, if it hasn't already |
| 09:23:46 | flyx | hm. the standard library looks pretty vast. is just everything someone writes that looks usable included there? |
| 09:24:16 | Araq | depends on what you're looking at, 0.9.2 was prior nimrod got a package manager |
| 09:24:38 | * | flyx is looking at http://nimrod-lang.org/lib.html |
| 09:24:58 | Araq | now that this exists, things are moved from the core to external packages |
| 09:25:27 | flyx | ah, okay |
| 09:26:35 | Araq | you're looking at the documentation for 0.9.2 fyi |
| 09:27:18 | Araq | but yes, the stdlib is quite big |
| 09:27:36 | Araq | it still has some serious holes like a lack of a bignum implementation |
| 09:28:45 | flyx | the stdlib may be a good reason to switch to nimrod |
| 09:33:14 | NimBot | Araq/Nimrod devel 274aff5 EXetoC [+0 ±1 -0]: Better names. |
| 09:33:14 | NimBot | Araq/Nimrod devel ec79e37 EXetoC [+0 ±1 -0]: Update test with new names. |
| 09:33:14 | NimBot | Araq/Nimrod devel 70d3078 EXetoC [+0 ±1 -0]: Use separate names for type-based allocation procs rather than overloading. |
| 09:33:14 | NimBot | Araq/Nimrod devel eb96089 EXetoC [+0 ±1 -0]: Update unit test. |
| 09:33:14 | NimBot | 4 more commits. |
| 09:34:34 | Araq | dom96: if you want the IpAddress stuff in net.nim, merge PR #972 |
| 09:34:48 | Araq | I won't because I'm not sure it belongs in there |
| 09:35:23 | Araq | EXetoC: can you move the mongodb wrapper to its own package? |
| 09:46:40 | NimBot | Araq/Nimrod devel 5626845 Miguel [+0 ±1 -0]: 'Connection' argument in 'Open' was not used, so MySQL host and port were always default ones. Now 'Connection' is treated as 'host:port'. |
| 09:46:40 | NimBot | Araq/Nimrod devel 0a322a3 Miguel [+522 ±210 -475]: Merge branch 'devel' of git://github.com/Araq/Nimrod |
| 09:46:40 | NimBot | Araq/Nimrod devel ff5d66a Miguel [+12 ±88 -3]: Merge branch 'devel' of git://github.com/Araq/Nimrod |
| 09:46:40 | NimBot | Araq/Nimrod devel 3963987 Miguel [+1 ±5 -1]: Merge branch 'devel' of git://github.com/Araq/Nimrod |
| 09:46:40 | NimBot | 2 more commits. |
| 09:46:57 | * | aftershave_ quit (Quit: Textual IRC Client: www.textualapp.com) |
| 09:47:18 | * | aftershave_ joined #nimrod |
| 09:47:28 | * | aftershave_ quit (Client Quit) |
| 09:47:43 | * | aftershave joined #nimrod |
| 09:51:41 | * | cark quit (Read error: Connection reset by peer) |
| 09:52:22 | EXetoC | Araq: I have. it's incomplete though, but the old one doesn't even work so I might as well remove it then I guess |
| 09:53:05 | Araq | the old one *used* to work, I tested it |
| 09:53:40 | Araq | but API stability is not something important for hippster nosql programmers |
| 09:54:17 | * | cark joined #nimrod |
| 10:00:53 | EXetoC | I know, you said that |
| 10:04:10 | Araq | EXetoC: ok, I'll simply remove the mongodb wrapper then, good? |
| 10:04:11 | * | nande quit (Read error: Connection reset by peer) |
| 10:07:18 | EXetoC | Araq: ok |
| 10:07:41 | EXetoC | no one is using it is what I'm saying, so that's why it can be removed now |
| 10:08:01 | EXetoC | unless someone happens to use it in conjunction with an old version of MongoDB, but that seems unlikely |
| 10:11:17 | EXetoC | so, should my xml example parse correctly? because a few characters outside the string literal are included in that case. in my case, the web server that I queried gave me a similar string with a raw ampersand in it |
| 10:22:22 | dom96 | Araq: If you just blindly execute commands then you won't get anywhere. |
| 10:22:53 | EXetoC | eventually he will |
| 10:23:33 | Araq | dom96: that's my point, I want to be able to do this |
| 10:23:57 | EXetoC | when? |
| 10:24:14 | dom96 | Araq: babel is not a mind reader |
| 10:24:36 | Araq | dom96: question is: do you mind patches that make it read my mind at least? |
| 10:24:57 | dom96 | Araq: Depends on the specific patch |
| 10:25:41 | Araq | babel update # should update all installed packages. do you agree? |
| 10:26:09 | dom96 | no |
| 10:26:34 | EXetoC | I think there should be a separate command for that |
| 10:26:36 | EXetoC | upgrade? |
| 10:26:58 | dom96 | Yes. 'update' already has a purpose. |
| 10:27:05 | dom96 | If you read --help you would know what it is. |
| 10:27:08 | * | filwit joined #nimrod |
| 10:27:11 | Araq | EXetoC: but this way lies madness, upgrade vs update are horrible for a non-native speaker at least |
| 10:27:52 | dom96 | Yeah. That's true. But it's either that or we will have: update-packages-json and update-packages |
| 10:28:29 | Araq | update-list and update-packages is good |
| 10:28:35 | * | q66 joined #nimrod |
| 10:28:48 | dom96 | Alright |
| 10:29:00 | dom96 | I can deal with that. |
| 10:29:40 | EXetoC | hey, what happened to short names? :p I think we short have short names for everything though |
| 10:29:54 | EXetoC | short/long options |
| 10:29:56 | dom96 | However, due to the way that babel operates, updating all packages isn't exactly trivial |
| 10:30:17 | dom96 | or rather, it's trivial but it's not trivial to only update the packages which have changed |
| 10:30:19 | Araq | dom96: I can imagine but you can simply iterate over the babel directory |
| 10:30:33 | dom96 | because there is no way to tell if the git repo changed. |
| 10:30:37 | dom96 | remotely |
| 10:31:25 | Araq | yeah I can imagine |
| 10:31:57 | dom96 | So I likely won't implement that. |
| 10:32:21 | EXetoC | we can change that behavior later then, because at some point the whole system needs to change in order to accomodate a bigger repository |
| 10:32:39 | dom96 | Although you've hit an interesting bug. |
| 10:32:48 | dom96 | The dependencies may need updating. |
| 10:33:01 | dom96 | Even if they are already installed |
| 10:34:58 | dom96 | But that may just be an issue with aporia's babel file. |
| 10:35:07 | dom96 | It depends on gtk2 >= 1.0, not #head |
| 10:36:12 | NimBot | nimrod-code/Aporia master d3f49aa Dominik Picheta [+0 ±1 -0]: Now depends on gtk2#head |
| 10:36:47 | dom96 | Yep. Works now. |
| 10:37:42 | Araq | ah |
| 10:37:49 | Araq | I see, cool |
| 10:38:07 | Araq | so we already have proper dependency tracking |
| 10:38:31 | dom96 | I think so. |
| 10:38:36 | dom96 | Depends what 'proper' means |
| 10:41:26 | * | DAddYE quit (Remote host closed the connection) |
| 10:41:53 | * | DAddYE joined #nimrod |
| 10:43:42 | * | Matthias247 joined #nimrod |
| 10:45:28 | * | io2 joined #nimrod |
| 10:46:01 | * | DAddYE quit (Ping timeout: 244 seconds) |
| 10:47:41 | * | io2 quit (Client Quit) |
| 10:55:50 | * | io2 joined #nimrod |
| 11:04:16 | * | Matthias247 quit (Read error: Connection reset by peer) |
| 11:10:30 | Skrylar | how not to do a filing system: yamaha's list of voice patterns |
| 11:10:58 | Skrylar | make sure that all 3 bytes have completely nonsensical order and settings that way its completely aggravating to index |
| 11:15:21 | * | Endy joined #nimrod |
| 11:22:52 | * | noam__ is now known as noam |
| 11:42:26 | * | DAddYE joined #nimrod |
| 11:47:16 | * | DAddYE quit (Ping timeout: 265 seconds) |
| 11:57:09 | * | filwit quit (Quit: Leaving) |
| 12:33:59 | * | eigenlicht quit (Ping timeout: 240 seconds) |
| 12:42:38 | * | psquid quit (Ping timeout: 240 seconds) |
| 12:43:11 | * | psquid joined #nimrod |
| 12:49:18 | * | psquid quit (Quit: worok) |
| 13:29:30 | * | darkf quit (Quit: Leaving) |
| 13:29:37 | * | BitPuffi1 quit (Ping timeout: 240 seconds) |
| 14:07:21 | * | xenagi joined #nimrod |
| 14:32:57 | * | [1]Endy joined #nimrod |
| 14:35:20 | * | eigenlicht joined #nimrod |
| 14:36:55 | * | Endy quit (Ping timeout: 265 seconds) |
| 14:36:56 | * | [1]Endy is now known as Endy |
| 14:46:40 | * | flaviu joined #nimrod |
| 14:47:19 | * | BitPuffi1 joined #nimrod |
| 14:50:02 | * | BitPuffi1 quit (Client Quit) |
| 15:06:49 | * | silven joined #nimrod |
| 17:24:49 | * | woodcut quit (Remote host closed the connection) |
| 17:52:58 | * | Endy quit (Ping timeout: 240 seconds) |
| 18:03:45 | * | nande joined #nimrod |
| 18:23:30 | OrionPK | hrm |
| 18:23:30 | * | r0b1 joined #nimrod |
| 18:23:50 | OrionPK | what's new folks |
| 18:25:55 | flaviu | dom96: Just came across your closure macro, its going to be very useful, thanks |
| 18:41:43 | * | Matthias247 joined #nimrod |
| 18:54:29 | * | Demos joined #nimrod |
| 19:23:41 | Varriount | flaviu: dom96 has a closure macro? |
| 19:24:12 | flaviu | Varriount: https://gist.github.com/dom96/9155536 |
| 19:24:38 | flaviu | Really neat, it lets (x,y)=>body |
| 19:27:55 | Varriount | Kinda like python's lambdas |
| 19:28:34 | flaviu | Seems much closer to scala's |
| 19:30:27 | Varriount | Araq: Removed the debug code from my PR |
| 19:31:12 | Varriount | flaviu: We need a nimrod-snippets repo. Its code like this that's convenient, but not exactly sophisticated enough for a package. |
| 19:35:19 | * | askatasuna joined #nimrod |
| 19:38:23 | flaviu | Varriount: Good idea, but I think that particular snippet should be extended into a package for functional programming |
| 19:40:33 | Matthias247 | I think that is even so much helpful that it should go in the stdlib or even the language ;) |
| 19:54:45 | Araq | Matthias247: we decided to leave it as a macro to show off nimrod's power |
| 19:59:02 | flaviu | Araq: wouldn't hurt for it to be in the stdlib. I stumbled upon it coincidentally. |
| 20:00:58 | flaviu | It doesn't show anything off if no one shows it off |
| 20:02:03 | fowl | be nice if (x: type, y: type) => .. worked |
| 20:13:07 | Demos | fowl, do you have tests for that assimp wrapper? |
| 20:15:09 | fowl | Demos, no but i have an example usage: https://github.com/fowlmouth/nimlibs/blob/master/fowltek/musings/nim3d/lib/models.nim |
| 20:16:02 | Demos | does that run? |
| 20:16:51 | Demos | I ask because I think the TMesh structure is wrong |
| 20:16:59 | fowl | it used to |
| 20:17:24 | Demos | I am writing a test now, but wanted to know if it was working for you. Is that valgrind clean? |
| 20:18:40 | fowl | thAT was a translation of a tutorial iirc |
| 20:24:21 | fowl | Demos, i just looked at aiMesh vs TMesh, it looks fine |
| 20:24:49 | Demos | oh is TMesh.texCoords an array of four arrays, one for each texcoord |
| 20:25:04 | Demos | so like all the us...... all the vs ..... all the ws ..... |
| 20:25:56 | fowl | C_STRUCT aiVector3D* mTextureCoords[AI_MAX_NUMBER_OF_TEXTURECOORDS]; |
| 20:26:34 | fowl | its an array of pvector3d |
| 20:27:46 | Demos | right, but I think in this case the pvector3d is an array of lengh mNumVertieces |
| 20:44:12 | * | OrionPK quit (Ping timeout: 252 seconds) |
| 20:49:32 | fowl | Demos, what makes you think that |
| 20:49:59 | Demos | I just did not understand the docs correctly |
| 20:50:20 | Demos | it supports sever uv(w) coords per vertex |
| 20:51:57 | NimBot | Araq/Nimrod devel 34ddb30 Clay Sweetser [+0 ±1 -0]: Tester now outputs all the test results of the current test run to json. |
| 20:51:57 | NimBot | Araq/Nimrod devel 4c2fec8 Clay Sweetser [+0 ±1 -0]: Removed debug code |
| 20:51:57 | NimBot | Araq/Nimrod devel bf982e9 Andreas Rumpf [+0 ±1 -0]: Merge pull request #981 from Varriount/tester/enhanced-json-output... 2 more lines |
| 21:01:58 | * | zahary1 quit (Read error: Connection reset by peer) |
| 21:02:01 | Demos | it would seem that if you use {.push dynlib:Foo.} to apply the {.dynlib.} pragma to functions than --dynlibOverride fails to work |
| 21:02:03 | * | zahary joined #nimrod |
| 21:05:32 | Araq | Demos: knowing how it's implemented, that seems unlikely |
| 21:07:59 | Demos | welp when I move dynlib: from the {.push.} pragma to each function the compilation fails with --dynlibOverride (as it should) but when I do not the compile succeeds but the program fails to start because it can not find the library |
| 21:10:43 | Araq | ok please report it |
| 21:15:19 | * | nastyanst quit (Remote host closed the connection) |
| 21:18:14 | fowl | im thinking of combining the chipmunk, ODE and bullet wrappers into one package |
| 21:20:26 | Araq | fowl: many small packages are preferable over fewer bigger ones. |
| 21:20:42 | Araq | The reason? github language usage statistics. |
| 21:21:31 | Demos | Araq, it seems that although it now works with --dynlibOverride it still loads a dynlib |
| 21:21:40 | Demos | I think it may just be a windows "thing: |
| 21:25:56 | dom96 | flaviu: Matthias247: Varriount: the closure macro will hopefully make it into the stdlib once some of the issues are ironed out |
| 21:26:39 | dom96 | I'm also planning to add a proc type description macro which will allow: (string, int) -> string |
| 21:26:53 | dom96 | instead of proc (x: string, y: int): string |
| 21:26:57 | Demos | ARRRRRGH pkg-config is broken in msys2 |
| 21:27:20 | flaviu | dom96: You can do that? is there any documentation for this feature? |
| 21:27:59 | dom96 | flaviu: You will be able to once I create the macro which will allow it. |
| 21:28:43 | dom96 | <fowl> be nice if (x: type, y: type) => .. worked |
| 21:28:56 | dom96 | The macro supports that IIRC |
| 21:29:14 | flaviu | No, I mean for the macro. Is there any documentation for type macros? |
| 21:30:36 | dom96 | There is no special macro required for this. Just read the general docs. |
| 21:35:45 | * | Endy joined #nimrod |
| 21:43:59 | Demos | wow so assimp actually fails to build in release mode on windows |
| 21:44:04 | Demos | sorry in debug mode |
| 21:53:22 | * | OrionPK joined #nimrod |
| 21:59:13 | fowl | thats ok |
| 21:59:23 | fowl | resil fails to build for me but at least devil still works |
| 22:01:07 | * | Demos_ joined #nimrod |
| 22:03:59 | * | Endy quit (Ping timeout: 240 seconds) |
| 22:04:55 | * | r0b1 quit (Ping timeout: 244 seconds) |
| 22:17:38 | renesac | "The difference in default timer function is because on Windows, clock() has microsecond granularity but time()'s granularity is 1/60th of a second; on Unix, clock() has 1/100th of a second granularity and time() is much more precise. " |
| 22:18:02 | renesac | "On either platform, the default timer functions measure wall clock time, not the CPU time. This means that other processes running on the same computer may interfere with the timing. |
| 22:18:33 | renesac | <-- how nimrod cpuTime() works in regards to that? |
| 22:20:07 | renesac | what are the equivalents to python's time() and clock() on nimrod, or the functions in times work the same independent of the OS?? |
| 22:20:21 | Varriount | renesac: cpuTime is a different kind of api call for nearly all systems. It's usually a monotomic timer that starts either when the application starts, or when the first call is made. |
| 22:21:18 | renesac | yeah, but what about the precision of cpuTime()? |
| 22:21:30 | renesac | what is it in linux, and is it the same on windows? |
| 22:22:16 | Araq | iirc cpuTime() - cpuTime() produces a time diff in ms on every OS |
| 22:22:30 | renesac | " However, it may measure the real time instead (depending on the OS)" |
| 22:23:10 | renesac | it would be useful to know which os it measures the real time |
| 22:23:17 | Araq | I think it does *cpu* time on posix, real time on windows |
| 22:23:46 | renesac | hum |
| 22:23:56 | Varriount | Araq: Actually, looking at the source of times.nim, it just uses clock() in the standard c runtime |
| 22:24:05 | renesac | so cpuTime() is probably equivalent to python's clock() |
| 22:24:53 | renesac | so I probably want the time() equivalent on unix |
| 22:27:19 | Varriount | Hm.. cpuTime() should probably be altered to use getHighPerformanceClock() on windows. |
| 22:29:02 | Araq | Varriount: yeah do it |
| 22:29:46 | renesac | getHighPerformanceClock() returns cpu time or real time? |
| 22:30:16 | renesac | and it would be good to have an equivalent precision on unix, as it is now 1/100th second precison aparently... |
| 22:30:40 | * | renesac still trying to wrap my head around this timer zoo |
| 22:30:44 | renesac | *his |
| 22:30:46 | Varriount | renesac: http://msdn.microsoft.com/en-us/library/windows/desktop/ff553053(v=vs.85).aspx |
| 22:31:02 | Varriount | I don't know what you mean by 'cpu time' and 'real time' |
| 22:31:40 | Varriount | Do you mean, a monotomically increasing counter, and a system wall clock time? |
| 22:32:28 | renesac | if it counts time used by other processes that may have stopped your process, for example, it is 'real time' |
| 22:32:37 | Araq | cpu time stops and resumes when the OS schedules another process |
| 22:32:40 | renesac | if it only counts the cputime of your process, it is cpu time |
| 22:32:51 | renesac | yeah |
| 22:33:24 | renesac | I *think* that a function named cpuTime() should return the cpu time |
| 22:34:07 | * | r0b1 joined #nimrod |
| 22:34:28 | Varriount | renesac: http://msdn.microsoft.com/en-us/library/ms725473.aspx |
| 22:34:49 | Varriount | Sorry, I don't have a map. Let's hope no-one gets lost and eaten by a tiger. |
| 22:34:54 | Araq | and I *know* that was hard to achieve portably and so we decided it's better to have an approximation for cpuTime |
| 22:35:05 | Araq | otherwise you do: |
| 22:35:19 | Demos_ | yup the assimp wrapper is binary incompatible with assimp, will fix |
| 22:35:23 | Araq | when defined(cpuTime): cpuTime() else: clockTime() |
| 22:35:39 | Araq | which is hardly an improvement over cpuTime() |
| 22:36:08 | Demos_ | also I think it is relevant that microsoft's version of the C++ standard lib has a function called std::chrono::high_percision_timer::now() and it has a resolution of 8ms, this is supposed to be the highest percision timer |
| 22:37:31 | Varriount | Demos: That probably uses the QueryPerformanceCounter api call |
| 22:37:44 | Araq | ugh the WinAPI is so ugly. Why so many vowels ... |
| 22:37:57 | Araq | unix proved they are not necessary |
| 22:38:15 | Varriount | Araq: That's not necessarily a good thing. |
| 22:38:32 | Varriount | Both are at the ends of two extremes |
| 22:38:38 | Araq | code simply looks clnr ths wy |
| 22:39:47 | * | r0b1 quit (Ping timeout: 265 seconds) |
| 22:40:10 | Araq | Varriount: my humor is still subtle, I guess |
| 22:40:13 | renesac | http://msdn.microsoft.com/en-us/library/windows/desktop/dn553408 <-- there are quite a few calls that give you 10~16ms resolution on windows... |
| 22:40:57 | Varriount | renesac: You could use this - http://msdn.microsoft.com/en-us/library/windows/desktop/ms683237(v=vs.85).aspx |
| 22:41:08 | Varriount | All you have to do is get the main thread handle. |
| 22:41:31 | Varriount | Why exactly do you need this time information for? |
| 22:41:40 | Demos_ | you need it in games |
| 22:41:49 | renesac | timeit clone |
| 22:42:05 | renesac | I'm using cpuTime currently |
| 22:42:13 | Demos_ | although the really fast timers have some problems where they can sometimes be different depending on which CPU core you are getting information from |
| 22:42:22 | Varriount | renesac: Then I'll see about updating cpuTime() to use QueryPerformanceCounter |
| 22:42:24 | renesac | and by the way it works it don't need that much precision |
| 22:42:37 | Varriount | That is what such things are supposed to use. |
| 22:42:40 | renesac | but 1/60th second is pretty low precision... |
| 22:43:13 | renesac | milisecond precision would already be cool, and the more the better |
| 22:43:17 | Araq | renesac: in system/ there is a much better module |
| 22:43:22 | renesac | I'm on linux now, though |
| 22:43:35 | Araq | I didn't expose it because then I have to maintain it |
| 22:43:42 | renesac | Araq, I saw it, for the gc right? |
| 22:43:47 | Araq | right |
| 22:43:48 | renesac | yeah |
| 22:44:22 | Demos_ | wow the GC makes valgrind a sad panda |
| 22:44:28 | renesac | Varriount, you know about this: https://github.com/Araq/Nimrod/blob/devel/lib/system/timers.nim |
| 22:44:29 | renesac | ? |
| 22:44:49 | Varriount | renesac: Vaguely |
| 22:46:59 | renesac | what I would need is this getTicks exposed |
| 22:47:48 | renesac | and the '-' |
| 22:49:28 | Araq | renesac: make a proper module out of it, copy&paste is allowed |
| 22:49:43 | renesac | hum, ok |
| 22:49:56 | Araq | I like to shield the GC implementation from a stdlib module |
| 22:50:21 | Araq | that will get overhauled soon because it doesn't follow "software engineering practice as introduced by Ruby" |
| 22:51:27 | * | Araq 's mood is on a summit |
| 22:55:06 | Varriount | Araq: Ooh, is it on one of the mountains in Garmisch? |
| 22:55:58 | * | r0b1 joined #nimrod |
| 23:00:36 | Varriount | Hi r0b1 |
| 23:05:06 | Araq | Varriount: do you like German music? |
| 23:06:26 | r0b1 | yo Varriount |
| 23:06:52 | Varriount | Araq: I haven't had much chance to listen to German music. The only one I know well is '99 Luftballoons' |
| 23:07:23 | * | dom96 <3 german music |
| 23:08:18 | Araq | hum who was the guy who used to live in germany? Demos? |
| 23:08:30 | Demos | nope |
| 23:08:51 | Varriount | I used to live in Germany |
| 23:09:04 | Varriount | However I went to school on a US military base. |
| 23:09:14 | Araq | ah good, so I'm not insane |
| 23:09:26 | * | Araq laughs and coughs |
| 23:11:16 | Varriount | "But I don't want to go among mad people." "Oh, you can't help that. We're all mad here. I'm mad. You're mad." |
| 23:11:49 | Varriount | Alice in Wonderland in a really strange story. |
| 23:12:12 | renesac | http://nadeausoftware.com/articles/2012/03/c_c_tip_how_measure_cpu_time_benchmarking <-- ok, this sorts out the cpuTime() confusion... with lots of #ifdefs |
| 23:13:04 | * | Araq played Alice in Wonderland on Hard |
| 23:13:12 | renesac | basically, for windows it should be 'GetProcessTimes' |
| 23:13:38 | * | Demos_ quit (Ping timeout: 244 seconds) |
| 23:13:46 | renesac | for other unixes, getrusage() seems good for microsecond precision |
| 23:14:14 | Araq | what is "other unixes"? |
| 23:14:34 | renesac | actually, for any unix |
| 23:14:38 | renesac | including linux |
| 23:14:46 | renesac | posix compilant oses |
| 23:14:52 | renesac | I mean |
| 23:15:40 | renesac | well, it describes each option, and offers a simple table as summary |
| 23:15:46 | renesac | in that link |
| 23:16:33 | Araq | muhaha, posix compliant means little |
| 23:17:12 | * | zezba9000 joined #nimrod |
| 23:17:22 | Araq | macosx will have its own way, BSD will follow an older version of the standard and linux only supports it with kernel 2.7.2 |
| 23:17:34 | * | carum joined #nimrod |
| 23:17:41 | renesac | yeah, clock_gettime( ) shows that... |
| 23:18:28 | renesac | but getrusage() seems consistent across all posix compilant systems |
| 23:18:54 | renesac | except for that 32 bit vs 64bit thing... |
| 23:19:12 | renesac | he don't specify when the 32bit interpretation was deprecated... |
| 23:19:46 | renesac | but I guess it is not a problem for anybody running an updated OS now |
| 23:19:53 | Araq | just make it work on haiku OS |
| 23:20:21 | renesac | he don't say anything about haiku OS there ^^' |
| 23:20:37 | Demos | was linux kernel 2.7.2 when relatime became default? |
| 23:20:45 | renesac | probably an "else: clock()" takes care of that |
| 23:20:46 | dom96 | Make sure it works in nimkernel too. You may need to implement most of the kernel first though :P |
| 23:21:07 | Araq | Demos: don't take me serious when i'm in the "mood" |
| 23:21:41 | Demos | yeah that is not a real kernel... :\ |
| 23:21:45 | Demos | hehe |
| 23:22:23 | dom96 | hrm... I should add some ASCII porn to nimkernel. That's all people want nowadays anyway. |
| 23:22:54 | Varriount | What's up with the Haiku OS mentions? I've heard it mentions a couple times on IRC, as well as in the forums |
| 23:23:33 | fowl | Varriount, haiku is the future |
| 23:23:58 | dom96 | I thought linux was the future |
| 23:24:06 | flaviu | fowl: how so? |
| 23:24:19 | Varriount | dom96: Use this -> http://text-mode.tumblr.com/post/55332433389/lesson-learned-dont-use-irc-for-presentations |
| 23:24:27 | * | Demos_ joined #nimrod |
| 23:24:32 | flaviu | dom96: 2014 is the year of the linux desktop :P |
| 23:24:37 | dom96 | Varriount: hah |
| 23:25:54 | Araq | Varriount: I like the Haiku project because it's not unix and tries to produce a decent desktop OS |
| 23:27:26 | dom96 | Meh. It's not written in Nimrod and therefore inferior in comparison to nimkernel. |
| 23:27:43 | EXetoC | flaviu: no 2015 |
| 23:27:49 | Araq | nimkernel is in a class of its own, ofc |
| 23:28:49 | dom96 | Nimkernel is a state of the art OS. ASCII, yellow background, text printing with a different background /and/ foreground color, and a choice of 16 fine colors. |
| 23:29:05 | dom96 | It doesn't get better than that. |
| 23:30:12 | Araq | also nimkernel introduces the revolutionary "everything is not implemented" abstraction |
| 23:30:30 | Araq | which ensures perfect interoperability between subsystems |
| 23:30:38 | dom96 | haha |
| 23:30:45 | dom96 | Precisely. |
| 23:31:14 | renesac | does it fit on a floppy disk? |
| 23:31:53 | Araq | nimkernel is a time saver par excellence, it lets you forget about your computer and go swimming instead |
| 23:32:38 | Araq | you'll never be more productive with a different OS |
| 23:33:04 | dom96 | It's so small that you can fit it on a single punch card! |
| 23:33:45 | flaviu | renesac: I think that it would with dead code elimination |
| 23:34:06 | flaviu | also, there are apparently 200MB floppy drives |
| 23:34:28 | renesac | those are zip drives |
| 23:34:54 | flaviu | renesac: really? looks floppy to me: http://www.karbosguide.com/images/hifd.jpg :P |
| 23:35:31 | renesac | oh, I haven't seen those |
| 23:38:37 | flaviu | dom96: error handling is broken, what if two errors happen right after each other? Wouldn't one be replaced before the user could read it? :) |
| 23:42:00 | dom96 | That is a very good point. |
| 23:42:31 | dom96 | A fix will come in Nimkernel 2.0. |
| 23:42:55 | dom96 | Yours for only $29.99! |
| 23:53:24 | Demos_ | is there a way to specify the packing of a struct |
| 23:53:33 | Demos_ | I am wrapping a library that uses #pragma pack(1) |
| 23:53:50 | Araq | Demos_: that's a tough one |
| 23:54:08 | Araq | you can import the header |
| 23:54:16 | Araq | (.header pragma) |
| 23:55:07 | Demos_ | not really ideal |
| 23:55:53 | Araq | depending on the struct, the pragma pack(1) might not be necessary |
| 23:56:34 | Demos_ | actually it may all be windows specific code |
| 23:57:44 | Demos_ | oh no it looks like these are types from the windows headers that this library just decided to copy |
| 23:58:44 | zezba9000 | Hey I'm dabbiling in the idea of making a C# to nimrod converter. |
| 23:58:45 | zezba9000 | Can't spend to much time in it yet, but find the idea promising. |
| 23:58:45 | zezba9000 | I can use NRefactory to convert the C# sytax tree to pure nimrod code. |
| 23:58:45 | zezba9000 | This would keep human readability from C# to Nimrod and make debuging easy when writing the converter. |
| 23:58:45 | zezba9000 | I plan on looking into this for personal reasons to get more performance out of C# and i'm sure others might find it usful if I end up doing it. |
| 23:58:46 | zezba9000 | Any thoughts or am I just speaking to the wrong crowd ;p |
| 23:59:07 | zezba9000 | ... Sorry for the word dump... |
