02:30 <KIMI> if i pass []const u8 string, and i store it in a struct variable same []const u8, what is size of struct? where string is stored?

02:34 <g-w1> iirc a slice is just the same thing as a `struct { ptr: [*]u8, len: usize }` so it would be size + size = 128 bits on 64 bit system

02:37 <KIMI> where it is stored? if i don't say [256] const u8, do i need to free memory?

02:40 <g-w1> a string literal ("this") is stored directly in the binary, and any other []const u8 you can choose how/where to store it. if you are getting it from an api, it should tell you if you own it now, and if you should free it

02:49 <KIMI> ok thanks

03:58 <moinstar> expected type '[]const u8', found '?[]const u8'

03:59 <moinstar> how to convert '?[]const u8' to '[]const u8'

04:46 <daurnimator> moinstar: `orelse` ?

06:27 <KIMI> is there a good library for parsing text line by line?

06:27 <KIMI> or an API

06:58 <daurnimator> KIMI: split into lines (with e.g. `.readUntilDelimiter*`) then write a parser for whatever you need within the line

07:03 <KIMI> what is it from?

07:18 <daurnimator> KIMI: its part of the reader interface

17:40 <hlolli__> is there a way to print pointer address? I'm asking to debug a case where I'm not sure if an object is the same or different.

17:40 <g-w1> "{*}

17:41 <hlolli__> thanks g-w1 ! std.debug.print("ptr={*}\n", .{ptr}); from the docs :)

19:49 <hlolli__> Are the std.log loggers made to be eliminated when compiling to an optimized binary?

19:53 <ifreund> hlolli__: they can be

19:54 <ifreund> you can customize the behavior by defining your own log function in your root source file

19:54 <hlolli__> it seems to be default tough?

19:54 <ifreund> indeed

19:55 <ifreund> they aren't all eliminated though, only the lower priority ones

19:55 <hlolli__> ok then I don't worry :) I use these log calls extensively, I was under the impression that wasm32-wasi was broken with optimizations, but when I also encountered it on x86_32-linux, I realized it must be removed.

19:55 <hlolli__> std.log.info to be percise, low priority indeed

19:57 <ifreund> you can override this behavior by setting the log level manually instead of letting the std determine it based off of build mode

19:57 <hlolli__> another question since you mentioned root source file, I've noticed that it's bit tricky to test files in isolation. If I want to test a nested file which imports file in ancestor directory, I'll get an error. Which I'm able to bypass with aliases. I don't know a good workflow to solve this, but I can solve it with --pkg-begin and --pkg-end statements

20:00 <hlolli__> in context of std.log, I'd love to create a namespace-ish file for my own loggers, but I'm afraid it would be a hassle to import them if I'm frequently changing from runnin build-exe, build.zig and zig test interchangably.

20:00 <ifreund> I'd recommend using a file in the toplevel directory of your project as a root test file and importing the others from their

20:00 <ifreund> you can add a test step to your build.zig

20:01 <ifreund> once you have a build.zig set up you should only need zig build really

20:01 <hlolli__> true, but I'd need to create build step for each and every file if I'd want to run one file at a time?

20:02 <hlolli__> but toplevel file for testing, which I'd comment out imports could work well actually.

20:03 <ifreund> you can also have build options which cause different things to be tested

20:03 <ifreund> build.zig is normal, turing complete zig code, anything is possible

20:04 <hlolli__> yes, I guess someone needs to create a nice testing framework with file watchers, aliases and glob file selectors.

20:04 <hlolli__> :) all good

20:05 <ifreund> that's certainly possible, just waiting for someone to do it :P

21:49 <andrewrk> it's designed to be declarative though. there are things that are more and less portable to do inside a build.zig

21:49 <andrewrk> this will become more obvious during the release cycle where we work on the package manager

21:59 <g-w1> is there any reason why stage2 relies on the system standard library instead of the one already in the zig repo? I have been having trouble with wrong versions of the standard library when developing stage2, but don't want to build the compiler release every time the repo updates. I think a good solution to this would just be to override std to be lib/std/std.zig if this is possible for stage2. or am

21:59 <g-w1> I missing something?

22:07 <andrewrk> g-w1, zig looks for the lib/ files relative to its own binary path, starting in the same directory and then searching upwards at parent directories

22:07 <andrewrk> could it be that zig build installed the lib/ files into the build/ directory next to the zig binary, and those are the ones being detected?

22:07 <andrewrk> you can run `zig env` to print the detected lib/ directory

22:11 <g-w1> I think my problem is that the version of zig that I am using is newer than the version of the compiler I am working on so the standard library is mismatched. My question was why doesn't stage2 just use the standard library in the zig repo? IMO this would make things simpler. a way to fix this would be for me to build zig myself, but would have to manage different zig versions that I built based on

22:11 <g-w1> different branches I am working on. I guess I could make a command that symlinks my global zig lib directory to the one in the zig repo, but I feel like it would be overcomplicating things

22:13 <g-w1> is it possible to override the stdlib from a build.zig file?

22:17 <g-w1> I am using a zig binary download from the website, but want to use the standard library in the repo for stage2. I realize that it may not have been clear before

22:18 <andrewrk> what is stage2 doing that is not using the standard library in the zig repo?

22:19 <andrewrk> s/that is not/instead of/

22:20 <g-w1> expected type 'std.builtin.Signedness', found bool; no member named 'is_signed' in struct 'std.builtin.Int'

22:21 <andrewrk> I mean, where is that std lib that is not the one in the zig repo?

22:21 <andrewrk> you shouldn't have to do this, I recommend *not* doing this: in build.zig: foo.override_lib_dir = "other/path/lib";

22:22 <andrewrk> you need to use a lib dir that matches the zig compiler. I'm confused how you got into this situation

22:22 <andrewrk> do you just need to run `make`?

22:22 <g-w1> ok. I guess I can just rebase with master every so often if this happens.

22:24 <andrewrk> I don't understand - how did this happen?

22:25 <andrewrk> when you update the std lib you have to rebuild the compiler

22:26 <g-w1> I tried to use the master compiler and stdlib from the website from today to compile stage2 from 2 days ago. should I always be doing ./build/zig build instead of zig build

22:28 <g-w1> I just rebased and it worked :). my lesson: make sure everything is up to date. thanks

22:53 <andrewrk> np