11:11 <unique_id> I just rewrote a simple file merging application in Zig (from C++). Less than 100 lines. Comparing the difference is a lot of fun. I'm much happier with the Zig code.

11:14 <unique_id> There's a lot of possible i/o errors and the Zig code just looks more robust to me and I know what it's doing.

11:26 <c15ade4> :)

11:31 <redj> unique_id: you can't show this C++ vs Zig for all to see?

11:40 <unique_id> I'll do it next time, sorry. I find that the explicit try statements make things much clearer. I also found a great use for orelse.

12:00 <unique_id> I'm building a zig library and then I'm going to depend on it from another zig project. I've never done this before, I don't know what to do.

12:01 <unique_id> the zig library has an index.zig file. From the main project I'd like to: @import("Library"), in the same way that you do @import("std")

12:44 <unique_id> I figured it out. Packages. exe.addPackagePath

20:01 <srgpqt> Hi all. Trying to use a c library from zig and can't seem to figure out how to deal with pointers to opaque types. The lib has a function that returns a pointer to an opaque type, and another function that accepts that pointer. When i try pass the pointer to that function, the compiler tells me "error: unknown-length pointer to opaque"

20:02 <andrewrk> hi srgpqt. here's an example of what's not allowed: [*]c_void

20:02 <andrewrk> what is allowed: *c_void

20:03 <andrewrk> you may also be interested in https://github.com/ziglang/zig/issues/1059

20:03 <srgpqt> yeah i already read through that issue, thanks

20:04 <andrewrk> I'm happy to help further if I can see some code snippets and error messages

20:05 <andrewrk> until #1059 is resolved, it might be worth your time to translate-c offline and save the results in a .zig file that you manually edit to fix up the pointer types

20:10 <srgpqt> still not clear exactly what syntax i need to make it work, perhaps i would have to manually fix up with translate-c? It's something like this:

20:10 <srgpqt> struct Base { int value; }

20:10 <srgpqt> typedef struct Foo Foo; // opaque type extending Base

20:10 <srgpqt> Foo *newFoo(void);

20:10 <srgpqt> void setBaseValue(Base *f, int value);

20:11 <srgpqt> so i want to call newFoo from zig, and pass that value to setBaseValue

20:16 <andrewrk> srgpqt, you'll need to @ptrCast(*Base, foo)

20:16 <andrewrk> it's unfortunate that setBaseValue has ?[*]T: pub extern fn setBaseValue(f: ?[*]struct_Base, value: c_int) void;

20:16 <andrewrk> better would be: pub extern fn setBaseValue(f: *struct_Base, value: c_int) void;

20:16 <andrewrk> this is what #1059 is talking about

20:20 <srgpqt> yeah

20:22 <srgpqt> i had already tried @ptrCast, it gives me "error: cast increases pointer alignment"

20:33 <srgpqt> I may have worked around it with some intermediate variables using @ptrCast and @alignCast... guess i'll find out

20:37 <andrewrk> srgpqt, ah ok that's the real issue - pointer alignment. I think you just have to settle for a @ptrCast and an @alignCast

20:37 <andrewrk> there's a better way to design a C API so that this does not happen

20:38 <srgpqt> For sure, but it would be nice if there was a convenient escape hatch for this use case. I'll try making a small helper function that does the casts for my situation

20:38 <srgpqt> Thanks

20:38 <andrewrk> agreed. #1059 describes that escape hatch