00:28 <Dirkson> How much tool support is considered acceptable for being able to write and maintain zig? Like, with C the minimum viable tool support is mainly "a text editor and a compiler". But if we end up doing implicit typing for both function arguments and function return,s it will more or less require the text editor to be able to find all libraries and understand all zig code in order to maintain non-trivial zig code.

00:46 <tiehuis> Dirkson: my view would be no special tool support is required

00:50 <Dirkson> That'd be my ideal too, but I'm curious if we have an official Word From DevPerson about this.

00:56 <tiehuis> i would expect andrewrk to think the same given his programming environment but he can say when hes back

02:43 <MajorLag> Currently I haven't personally had much difficulty with just a text editor and the compiler, I hope it remains that way. I see the point about inferred return types but I think the trait, or more like "userland typechecking", proposal can help with that. I always felt a bit dirty taking "var", but after implementing a form of userland traits and typeclasses I feel much more comfortable with it. Making those checks

02:43 <MajorLag> ould help.

02:45 <kristate> woke up 11:45 JST

02:52 <scientes> good morning kristate

02:52 <kristate> hi!

02:53 <kristate> what time is it there?

02:53 <scientes> 20:00

02:55 <kristate> not so bad -- how was your day?

02:55 <scientes> worked most of the day

02:55 <scientes> now i am wondering what I should work on programming-wise

02:59 <scientes> nice work on 1389

03:03 <kristate> thanks

03:03 <scientes> I feel i'm accumulating bugs in zig

03:03 <scientes> My cpp skills are not good enough to work on the compiler

03:03 <kristate> scientes: but you know C, right?

03:04 <scientes> yes

03:04 <scientes> i haven't looked at how much cpp is used

03:04 <kristate> C++ isn't so different -- the only thing you have to watch-out for are templates and decorators

03:05 <scientes> yeah but I look at simple C++ headers and my mind boggles, iterators for example

03:05 <kristate> exactly

03:05 <kristate> google's c++ guide is good. they don't allow most of their programmers to use all the featureset of c++

03:06 <scientes> although I do know enough to know that there are places where c can't be performant (but zig can) because of adherance to the C ABI

03:06 <kristate> I personally despise C++, but since LLVM is c++, it makes sense why andrewrk chose to use it

03:07 <kristate> the sooner we can move over to zig, the better

03:07 <scientes> go was implemented in C, and then they machine-translated it to go (but they never used macros)

03:07 <kristate> here is the guide: https://google.github.io/styleguide/cppguide.html

03:09 <scientes> I had big disagreement with some of the review of my code. like the suggestion to not return arrays

03:10 <scientes> but thats just another zig bug https://github.com/ziglang/zig/issues/863

03:10 <scientes> anyways....

03:11 <kristate> scientes: wonder what it would take to fix this

03:12 <kristate> it looks like #770 was fixed

03:56 <scientes> comptime has the potential (with lots of work) to be very powerful

03:57 <scientes> like regular express compilation and execution

03:57 <scientes> but I wonder if an existing regular expression implementation can be modified to support comptime (and would non-zig code be acceptable to be integrated?)

04:48 <tiehuis> scientes: pretty sure we would only need a comptime allocator for this to be work: https://github.com/ziglang/zig/issues/1291

06:21 <andrewrk> kristate, hi there

06:21 <kristate> andrewrk: hi

06:21 <kristate> how was your vacation?

06:22 <andrewrk> quite pleasant seeing my family. but now I'm ready to get back to work :)

06:22 <kristate> :)

06:23 <kristate> andrewrk: this BigInt stuff is crazy

06:23 <andrewrk> I'm going to focus on improving the stage1 compiler for the 0.3.0 release

06:23 <andrewrk> (starting with all the open PRs)

06:23 <kristate> andrewrk: thank-you!

06:23 <kristate> stage1 is not so bad -- I like it

06:32 <kristate> andrewrk: I am going to say something really weird -- but what if we start rewriting parts of stage1 in zig? kind of like a stage1.5

06:33 <kristate> andrewrk: BigInt needs a Big Overhaul

06:33 <andrewrk> I think that could make sense, as a transition. we will never lose stage1 though

06:35 <andrewrk> stage1 must be source, not output

06:35 <andrewrk> and it has to be able to build stage2 in 1 step

06:36 <kristate> andrewrk: yeah, more I think about it, doesn't make sense

06:37 <andrewrk> is it really that bad? I thought the problem was that it had a bunch of untested code paths

06:38 <kristate> twos_compliment is doing too much work

06:39 <kristate> we're only using BigInt in 87 places (outside of BigInt itself)

06:40 <andrewrk> I mean that logic has to go somewhere

06:40 <andrewrk> if you can refactor it to delete some lines of code, by all means, go for it

06:40 <kristate> andrewrk: yeah, about 2 hours into a patch

06:40 <andrewrk> cool

06:41 <andrewrk> I do need 1 more day of rest before I get back into it

06:41 <andrewrk> I'm about to get on a red eye flight

06:41 <andrewrk> talk to you later

06:42 <kristate> andrewrk: okay, sleep well! welcome back

18:35 <DutchGh0st> is there like a 'goto' concept in zig?

18:37 <scientes> DutchGh0st, yes, named breaks https://github.com/ziglang/zig/issues/630

18:39 <unique_id> I like goto myself, but I've not had a need for it in Zig because of 'named breaks'. Though in a rare case my code became a little bit more verbose.

18:39 <scientes> yeah i like both goto and VLAs and @alloca

18:40 <unique_id> What does the % do in Andrew's example on that page? I assume that syntax has been removed.

18:40 <DutchGh0st> a 'named' break just puts a lable on a loop-struct, and you break anywhere out of any loop?

18:40 <scientes> errdefer

18:40 <DutchGh0st> or am I completely wrong?

18:40 <scientes> yeah there was a de-mysterious-symbols movement in zig

18:41 <DutchGh0st> is it more or less equal to the labeled breaks in Rust? or more powerfull?

18:45 <unique_id> I don't trust myself to answer questions on breaks, but the docs have some info and there's this: https://github.com/ziglang/zig/issues/1327

18:50 <DutchGh0st> ah mhhh

18:50 <DutchGh0st> how is your game going btw??

18:58 <unique_id> I've been in learning mode for the last few days. I've been seeing how today's games implement particle systems, decals, etc. It's looking like my game will be going with what's called a tiled forward renderer. Basically you split your screen into tiles and use a compute shader to collect a list of things per tile. Like a list of lights, a list of decals, a list of particles. This makes things more efficient and it's a forward

18:58 <unique_id> shader which means I get MSAA.

19:01 <unique_id> Basically I'll be going with an architecture similar to what Doom did. But my game won't actually be heavy on graphics requirements though it will require a vulkan compatible card.

19:03 <unique_id> 90% of steam users have DX12 supported card by now (a bit less for vulkan) and it'll increase rather quick. It'll definitely leave some people out sadly, but w/e.

19:08 <unique_id> (i don't know how many of those dx12 users are chinese playing through internet cafes....could be a lot)

19:12 <unique_id> I really need decals, I can't get them easily without tiled forward or deferred :)

19:21 <DutchGh0st> mhhh

19:22 <unique_id> DutchGh0st: decals are anything that's layed out directly onto anything without adding height. Text (possibly), blood splatter, explosion damage, etc

19:23 <DutchGh0st> ohhh I see

19:23 <DutchGh0st> seems all difficult to get good at haha

19:25 <unique_id> yeah I'm taking on a lot of roles that people usually specialize at. It takes a long time of being very confused until things start to make sense

19:38 <unique_id> DutchGh0st: I started reading about deferred shading 2 years ago, it was only recently that I began to understand them. I could of course just use a ready-made game engine but had I done that I'd be a skilless game developer with a tool much too powerful for me. And I'd pick the same strategies that the largest studios employ. I'd depend on an animator for animations instead of learning math tricks to reduce the number of

19:38 <unique_id> keyframes required, allowing programmer animations :)

19:39 <unique_id> So for me picking a game engine is no shortcut. Learning how everything works is inevitably going to make me a better developer, with a much better understanding of what's possible

19:39 <DutchGh0st> I remember I had to make a game for highschool

19:39 <DutchGh0st> we had like 3 teams

19:39 <DutchGh0st> the animators or whatever, the map creators, and the programmers

19:39 <DutchGh0st> note: We had never programmed anything before

19:40 <scientes> is there some sample code that uses the new tcp stuff?

19:41 <DutchGh0st> and like after a while the animators and map creators had like all these maps an animations and whatnot.....and it was just a hard job for me and the rest of the programmers to put everything together well

19:42 <unique_id> yeah that's understandable. They're creating everything in isolation, in their little tools. The whole game is then left

19:42 <DutchGh0st> I wonder what we could have produced if we'd gotten propper lessons

19:44 <DutchGh0st> cuz it at the start of the year it was like 'youuu gotta make 2 games. They've to be finished in like 6 months. Goodluck'

19:57 <scientes> thats horrible teaching

19:57 <scientes> you need smaller projects than that

20:12 <DutchGh0st> yes!

20:12 <DutchGh0st> we had 2 crappy games at the end

20:12 <scientes> like my command-not-found implementation https://github.com/shawnl/command-not-found is even complicated enought that porting it to zig would be significant (/etc/groups support, gettext .posupport)

22:11 <scientes> how do i implement non-blocking io with async?

22:18 <scientes> i'm getting error.WouldBlock when i try to read

22:20 <scientes> but its in a async thread, I want it to block

23:00 <scientes> I need an example of how to use EventFd

23:00 <scientes> there are no tests

23:01 <scientes> I don't know how to call read with event.loop

23:35 <scientes> and why is it pread with a v--vector?