Null pointer optimization when the pointer is in a structure

[wooster0]

comptime {
    @compileLog(@sizeOf(?*u8));
}

Compile Log Output:
@as(comptime_int, 8)

Because this pointer doesn't have the allowzero attribute it, as expected, Zig uses address zero for null.

However when you do this:

const Thing = struct {
    pointer: *u8,
};

comptime {
    @compileLog(@sizeOf(?Thing));
}

Compile Log Output:
@as(comptime_int, 16)

That doesn't apply anymore.

Additionally, in more complicated cases like this:

const Thing = struct {
    hello: u64,
    world: enum { a, b, c },
    pointer: *u8,
};

comptime {
    @compileLog(@sizeOf(?Thing));
}

It should be able to figure out that it can use any pointer's zero address within the structure to represent null for ?Thing. That would make the size shrink to 24 bytes from 32 currently.
This could be combined with #168 to move non-allowzero pointers to the top so it can perhaps check for null a bit faster? I don't know.

[mlugg]

Duplicate of #104.

Do not file a proposal to change the language. You have already been given a temporary time-out from the issue tracker for ignoring this rule a few months ago. If you continue to disregard this clear rule, you will likely end up being permanently banned from the issue tracker.

[wooster0]

Woah, so that’s why I wasn’t able to do anything on this repository for a whole month. Never saw that comment. Crazy for someone that occasionally contributes through PRs and reports issues but I guess no one cares.

Also didn’t see this as a language change when writing this. Literally just an optimization that doesn’t add any feature observable through anything other than the return value of @sizeOf changing.

Also guess I would like to know why I’m instantly given a 1 month ban without any warning. Could’ve literally just said “we’re going to ban you” or something instead of that

[wooster0]

I never saw that comment so I thought maybe GitHub is just buggy or something or I’m shadow banned for some reason which is really painful when you don’t even know the cause. Like just take one second to search if that person isn’t actually a contributor or something to at least have the mercy to give them a simple warning instead of that. Can’t comprehend that decision.

Also doesn’t matter but there were tons of language proposals still being accepted by non-core team members when that rule was introduced so the rule wasn’t really followed initially so I thought oh maybe I’m going to be lucky again and they’re not gonna close this

[mrjbq7]

Crazy for someone that occasionally contributes through PRs and reports issues but I guess no one cares.

Wow, 142 closed PR's, many of them merged.

[wooster0]

Also, what is that, leaving someone confused for a whole month? How about just a week? Or even just a day? Or simply issuing a warning?

[mlugg]

Contributing to the issue tracker or the repository does not make you exempt from its rules. Retroactively arguing that you shouldn't have been punished (or should have been punished less) for knowingly disregarding the language proposals rule is... not a great look.

Please drop this discussion now.