slicing sentinel-terminated array or pointer with .. as end index should give a sentinel-terminated slice

[andrewrk]

Follow-up from #3728. Depends on #863.

test "slice to the end of a null terminated pointer" {
    const string_literal = "hello, world";
    comptime expect(@typeOf(string_literal[5..]) == *const [7:0]u8);
    const slice: [:0]const u8 = string_literal;
    comptime expect(@typeOf(slice[5..]) == [*:0]const u8);
}

[N00byEdge]

Should it? ptr[std.mem.len(ptr) + 1..] could definitely be an unterminated pointer, while ptr[std.mem.len(ptr) + 1..:0] clearly is terminated, and only two characters longer to write.

[mikdusan]

wait, title asks for sentinel-terminated slice
but the OP example tests for a sentinel-terminated pointer
and current master produces a sentinel-terminated slice

so.... maybe OP example actually intended to test for sentinel-terminated slice and master is already there?

const std = @import("std");

pub fn main() void {
    const string_literal = "hello, world";
    //comptime expect(@typeOf(string_literal[5..]) == *const [7:0]u8);
    std.log.debug("A: {}", .{@TypeOf(string_literal)});

    const slice: [:0]const u8 = string_literal;
    //comptime expect(@typeOf(slice[5..]) == [*:0]const u8);
    std.log.debug("B: {}", .{@TypeOf(slice)});

    std.log.debug("C: {}", .{@TypeOf(slice[5..])});

    rtime(slice);
}

fn rtime(s: [:0]const u8) void {
    std.log.debug("D: {}", .{@TypeOf(s[5..])});
}

output:

debug: A: *const [12:0]u8
debug: B: [:0]const u8
debug: C: *const [7:0]u8
debug: D: [:0]const u8

[mlugg]

Yeah, the title of this issue seems a little confused but it definitely seems like what it proposes is implemented and has been for a while.

The second assertion in the OP fails because the slice is comptime-known so you get an array pointer. Plus, you'd never get a many-pointer from that operation; you'd get a slice. But you do indeed get a sentinel-terminated slice as was intended.