2 ms·

His issues (from 3.5 years ago): "- The string handling is braindead" How so? Not a lot of specificity here. "- The compiler and stdlib is immature. This see

His issues (from 3.5 years ago): "- The string handling is braindead" How so? Not a lot of specificity here. "- The compiler and stdlib is immature. This seems like the first programming language the guy has written. You should have a few failures to learn from before you make a Serious Programming Language imo." The Hare compiler isn't immature? As noted, this was 3.5 years ago. "- Code run at compile time is a neat idea but has a weird syntax and really really complicates the internals of the compiler and related tooling." And yet this has become one of Zig's strongest points.

dralley 5y ago
Maybe it has changed in the last 3 years, but "comptime" doesn't seem like such a "weird syntax" to me?
lhorie 5y ago
Re: strings, this is a thread about that topic: https://github.com/ziglang/zig/issues/234 https://github.com/ziglang/zig/issues/234 It strikes me as a difference in opinion wrt at which level should strings be handled (zig folks saying it should be a userspace/stdlib concern, ddevault saying it should be a core language concern). IMHO the point of contention seems to be that ddevault considers stdlib an integral part of a language, whereas zig has an ideology that aggressive support for freestanding, no-stdlib-whatsoever workflows is part of its scope. From my understanding from the other HN thread, Hare appears to use QBE as a backend, which only targets x64 (whereas Zig says it'll continue to use LLVM for prod builds). Zig is also somewhat unique among low level languages in that it put quite a non-trivial amount of legwork on cross-compilation plumbing (e.g. the article about compiling C extensions for go via zig[0]) [0] https://dev.to/kristoff/zig-makes-go-cross-compilation-just-work-29ho https://dev.to/kristoff/zig-makes-go-cross-compilation-just-...
robinei 5y ago
Since then Zig has come quite far in its self-hosted rewrite. So the point of it being Andrew's first compiler is no longer valid.

dralley 5y ago

Maybe it has changed in the last 3 years, but "comptime" doesn't seem like such a "weird syntax" to me?

lhorie 5y ago

Re: strings, this is a thread about that topic: https://github.com/ziglang/zig/issues/234 https://github.com/ziglang/zig/issues/234 It strikes me as a difference in opinion wrt at which level should strings be handled (zig folks saying it should be a userspace/stdlib concern, ddevault saying it should be a core language concern). IMHO the point of contention seems to be that ddevault considers stdlib an integral part of a language, whereas zig has an ideology that aggressive support for freestanding, no-stdlib-whatsoever workflows is part of its scope. From my understanding from the other HN thread, Hare appears to use QBE as a backend, which only targets x64 (whereas Zig says it'll continue to use LLVM for prod builds). Zig is also somewhat unique among low level languages in that it put quite a non-trivial amount of legwork on cross-compilation plumbing (e.g. the article about compiling C extensions for go via zig[0]) [0] https://dev.to/kristoff/zig-makes-go-cross-compilation-just-work-29ho https://dev.to/kristoff/zig-makes-go-cross-compilation-just-...

robinei 5y ago

Since then Zig has come quite far in its self-hosted rewrite. So the point of it being Andrew's first compiler is no longer valid.