Add 'u' specifier to std.format #3970
Conversation
daurnimator
added
the
standard library
| output: fn (@TypeOf(context), []const u8) Errors!void, | ||
| ) Errors!void { | ||
| var buf: [4]u8 = undefined; | ||
| const len = std.unicode.utf8Encode(c, buf[0..]) catch unreachable; |
There was a problem hiding this comment.
why is this catch unreachable? what if someone e.g. passed in 0x1fffff
There was a problem hiding this comment.
attempt to unwrap error: CodepointTooLarge
Personally I'd prefer to have utf8EncodeUnsafe with 0xFFFD replacement.
There was a problem hiding this comment.
I'm OK with this function asserting that the codepoint fits, with a bit more defensive coding:
- Document the assertion in doc comments of the function
- Rather than
catch unreachable,catch |err| switch (err)and then list the error(s) explicitly that are unreachable. This will cause compile errors if new errors are added to the set.
| @compileError("Cannot print integer that is larger than 8 bits as a ascii"); | ||
| } | ||
| } else if (comptime std.mem.eql(u8, fmt, "u")) { | ||
| if (@TypeOf(int_value).bit_count <= 21) { |
There was a problem hiding this comment.
Does a comptime int pass this check?
There was a problem hiding this comment.
Copy-pasted from c modifier.
There was a problem hiding this comment.
I do want this to handle comptime_int please. The code for the c modifier is also incorrect if it cannot handle them. Until recently, due to limitations of the zig language, comptime_ints could not be passed to formatted printing. However that is now fixed, and comptime ints can be passed.
I would be OK with a non-explicit @compileError call here, allowing the @as to do the job. That would be my personal strategy to handle comptime_int correctly. Note also that the @as is redundant. Every parameter always gets type coerced to the parameter type.
andrewrk
left a comment
andrewrk
left a comment
There was a problem hiding this comment.
The API and doc changes are good 👍
a couple improvements in the implementation and then this is merge-ready.
| @compileError("Cannot print integer that is larger than 8 bits as a ascii"); | ||
| } | ||
| } else if (comptime std.mem.eql(u8, fmt, "u")) { | ||
| if (@TypeOf(int_value).bit_count <= 21) { |
There was a problem hiding this comment.
I do want this to handle comptime_int please. The code for the c modifier is also incorrect if it cannot handle them. Until recently, due to limitations of the zig language, comptime_ints could not be passed to formatted printing. However that is now fixed, and comptime ints can be passed.
I would be OK with a non-explicit @compileError call here, allowing the @as to do the job. That would be my personal strategy to handle comptime_int correctly. Note also that the @as is redundant. Every parameter always gets type coerced to the parameter type.
| output: fn (@TypeOf(context), []const u8) Errors!void, | ||
| ) Errors!void { | ||
| var buf: [4]u8 = undefined; | ||
| const len = std.unicode.utf8Encode(c, buf[0..]) catch unreachable; |
There was a problem hiding this comment.
I'm OK with this function asserting that the codepoint fits, with a bit more defensive coding:
- Document the assertion in doc comments of the function
- Rather than
catch unreachable,catch |err| switch (err)and then list the error(s) explicitly that are unreachable. This will cause compile errors if new errors are added to the set.
|
Works for me as is. You got an idea. |
3 participants
Outputs integer as an UTF-8 sequence.