00:15 <jzelinskie> o/

09:15 <jfo> \o

12:06 <MajorLag> Anyone want to save me some time and explain/link to an explanation of what libicuuc is and why it is a dependency if I build Zig on Ubuntu Artful but not on Debian Jessie? Is it necessary in any way? because it adds about 30MB.

16:06 <sand-witch> Hello! I just want to express my appreciation here for those working on this language. I'm incredibly impressed!

16:07 <sand-witch> I have a fun quote from Jon Carmack that I'd like to share, from his facebook: https://www.facebook.com/permalink.php?story_fbid=2110408722526967&id=100006735798590

16:08 <sand-witch> The quote is: I do have plenty of days where, like many older programmers, I think “Maybe C++ isn’t as much of a net positive as we assume...”.

16:08 <sand-witch> I think Zig might get a lot of C and C++ programmers interested

16:14 <sand-witch> I do have one question: Why can structs not be passed by value? I assume this is to get rid of the endless decisions on whether to pass by value or pointer, and what that even means depending on size of struct and calling convention?

16:15 <sand-witch> Basically, passing by value vs pointer doesn't mean anything anyway, so we might as well force a const pointer?

16:17 <sand-witch> The ABI then picks something efficient to do

16:29 <MajorLag> That's actually sort of what is being planned, as I understand it. See #733: https://github.com/zig-lang/zig/issues/733

16:32 <andrewrk> hi jzelinskie o/

17:17 <andrewrk> MajorLag, I don't know what libicuuc is for

17:17 <andrewrk> it's not a direct dependency of zig

17:17 <andrewrk> it's unfortunate how many dependencies llvm/clang are pulling in

17:23 <MajorLag> andrewrk: it's apparently some unicode things. libxml2 can be compiled against that or iconv. Ubuntu 17.10 built libxml2 against libicuuc, Debian Jessie built against iconv.

17:23 <andrewrk> oh wow

17:24 <andrewrk> this is unreasonably complicated

17:26 <andrewrk> goodness, libxml2 is 200,000 lines of C

17:27 <andrewrk> the entire zig compiler is 51,000

17:28 <andrewrk> maybe we could make an xml parser in zig exporting a C API and have LLD depend on that instead :)

17:37 <sand-witch> The compilers for Odin and Loci are a similar size to Zig. Kinda cool. Meanwhile over at Clang... (though not exactly a fair comparison)

17:40 <sand-witch> oh Jai is also a similar size IIRC

17:44 <sand-witch> By the way, It's quiet likely that sometime in the future once C++ modules are here, they'll introduce some kind of way to change how the language works on a per-module basis. This will do things like turn explicit on by default, that kind of thing. But my god, even then, the complexity isn't going to disappear from C++. You guys are saving me from an endless cycle of C++ induced depression.

17:45 <sand-witch> So thanks andrewrk :)

17:46 <andrewrk> sand-witch, are you working on some zig code?

17:51 <sand-witch> No, I just like zig the most of all the recent languages arriving. I'm actually having some trouble compiling the compiler, but I'll work on that a bit later. The errors are stuff like "error: ‘HashString’ is not a member of ‘llvm’" --- It seems deps/lld is reading from my /usr/local/include/llvm which is very recent. I deleted it but then I get different errors. Will troubleshoot later.

17:52 <andrewrk> ok, feel free to ask for help getting it working. is that ubuntu?

17:52 <sand-witch> yes, 17.10 I think

17:52 <MajorLag2> @andrewrk, that'd be nice, because I'm pretty sure I got it to build against musl and libc++ otherwise.

17:54 <MajorLag2> sand-witch: did you use the llvm-6.0 ppa?

17:54 <andrewrk> I've tested to make sure deps/lld picks up .h files from itself before system files

17:55 <sand-witch> MajorLag2: I compiled from master recently

17:55 <andrewrk> that's the problem

17:55 <andrewrk> zig master branch is tracking llvm 6 exactly. llvm master branch can break compatibility with every commit

17:56 <sand-witch> yeah but I deleted it so it'd use /usr/include but then I'm getting different errors. I'll figure it out :)

17:56 <sand-witch> it meaning /usr/local/include/llvm

17:57 <sand-witch> now it's probably a version that's too old

17:57 <MajorLag2> For the record, I didn't have any trouble using the llvm-6.0 ppa, clang, and making sure libxml2-dev was available.

17:57 <sand-witch> ok good to know!

17:59 <ltr_> i think the best way to test drive zig is with docker

17:59 <MajorLag2> I've potentially got an even easier way: I've built a self-contained AppDir of the zig compiler.

18:00 <MajorLag2> it's a bit hefty at 130MB, but now that I've excised libicuuc that'll drop a good 30MB

18:00 <andrewrk> MajorLag2, if you figure out how to automate that, we can put it in travis CI and provide it as an artifact

18:01 <MajorLag2> I actually have automated it. If I add my build script to a cron job, it'll just go ahead and build a new tar.bz2 every time master updates

18:02 <MajorLag2> and then it will upload it to my site and update the displayed build number and updated date too.

18:02 <andrewrk> ah that's nifty. mind if I link to that from ziglang.org/download?

18:03 <MajorLag2> Not at all, that was always the plan, I just wanted to do some more testing and trimming before I was going to bring it up.

18:03 <andrewrk> cool

18:07 <MajorLag2> Oh, I should add langref to it

18:08 <MajorLag2> I'm not sure how compatible my build process is with travis ci, but that sounds like a goal. I'll look into it.

18:08 <andrewrk> good idea. 57KB for up-to-date docs that work offline is a small price to pay

18:10 <MajorLag2> compressed tarball is 28MB, uncompressed dir is ~98MB (without libicuuc, next version won't have it), so 57KB isn't even going to be noticed. Considering I include literally every library it needs, being only about 40MB larger than the Windows version isn't bad.

18:10 <andrewrk> not bad at all. it could probably be smaller if we tried to statically link everything

18:10 <andrewrk> that can be tricky

18:16 <MajorLag2> Yeah, glibc is pretty actively hostile to the concept. that's why I was dabbling with musl/libc++. I got the embedded LLVM stuff built that way after some hacking, but ran into a lot of missing c++ std symbols so I took a break on it because I knew libxml2 would be trouble anyway.

18:18 <andrewrk> MajorLag2, have I told you about the Grand Bootstrapping Plan?

18:23 <andrewrk> I'm going to make an issue

18:26 <MajorLag2> Maybe, was that the one with making a zig C abi?

18:26 <andrewrk> it's related to that

18:40 <andrewrk> MajorLag2, https://github.com/zig-lang/zig/issues/853

20:02 <andrewrk> jacobdufault, how's it going over there? anything I can help with on my end?

20:27 <jacobdufault> been a bit busy these past few weekends, still slowly working on it

20:27 <jacobdufault> if you'd like to close the PR for now that's okay, I'll reopen when ready

20:28 <jacobdufault> always too many things to do :)

20:28 <andrewrk> I am all too familiar with that sentiment

20:30 <GitHub55> [zig] andrewrk closed pull request #824: [WIP/NOTREADY] WIP json metadata (master...wip-json-render) https://git.io/vAh5y

20:30 <sand-witch> I see so many commits @ LibreOffice that have to do with overflow. It just makes me think how awesome this language will be for big software projects.

20:42 <sand-witch> Sorry for being the weirdo cheerleader here, but one last thing. I've always hated the concept of library wrappers and the fact that C++ people wrap C libraries instead of introducing a defer keyword has always blown my mind. The C++ culture is broken in my opinion, and this language can wake a lot of people up.

20:43 <andrewrk> sand-witch, I look forward to seeing what you create

20:44 <sand-witch> I do too :D