Conversation
The comment contained the typical mistake os saying "safe" when "sound" is clearly meant (transmutes are never safe but may be sound), so this fixes it.
|
utACK a3b6640. Arguably the "Because" lines in the doccomments are unnecessary. From a user point of view, these conversions make sense because we've provided a safe method that provides them. The user doesn't care what representation magic is needed to get it to work. Tests would be nice, I guess. The code appears to be correct, though it's much more aggressively optimized than anybody else would do. CI is failing because of a clippy lint (looks like one of your pointer casts has a convenience function now) and also the API files, which if they annoy you too much, me or Tobin can add. |
Originally it was proposed that `AsRef<str> for [Char]` and other conversions are implemented. However these are disallowed by the orphan rules. The conversions are still desirable, so this commit adds them as "static" methods of `Char`. This specifically adds only conversions to slices since these are most common and arrays will coerce to them if needed.
`Char` should be very similar to `char` but it was missing `Display` impl. This commit adds it by casting `&char` to `&str` and calling that which gives us formatting options for free.
Naively derived `Debug` impl on `Char` would produce the character names
rather than the characters themselves like `char` does. To improve
consistency and brevity, this manually implements `Debug` by creating a
stack-allocated `'{x}'` string and formatting that one.
Kixunil
force-pushed
the
char-improvements
branch
from
a3b6640 to
2713e69
Compare
|
Addressed all the feedback, including adding tests. |
|
Can you run the formatter? |
| // Because all chars are ASCII a slice of chars is also guaranteed to be valid slice of | ||
| // bytes. |
There was a problem hiding this comment.
nit: I rekon its better without this comment. Mainly because 'all chars are ASCII' is not true if one interprets 'chars' as the Rust type. But also because the SOUNDNESS line already documents the code.
|
Nice PR. |
3 participants
These are a few follow ups to #236
I'm too tired for tests rn, will add them later.