Unicode handling in String and WString #362
Conversation
According to https://design.ros2.org/articles/wide_strings.html, the `string` and `wstring` types must use the UTF-8 and UTF-16 encodings (not necessarily enforced), cannot contain null bytes or words (enforced), and, when bounded, are measured in terms of bytes or words. Moreover, though the rosidl_runtime_c `U16String` type uses `uint_least16_t`, Rust guarantees the existence of a precise `u16` type, so we should use that instead of `ushort`, which isn't guaranteed to be the same as `uint_least16_t` either. (Rust doesn't support platforms where `uint_least16_t != uint16_t`.)
There was a problem hiding this comment.
Hey sorry for the long delay @nwn, these changes LGTM.
Rust doesn't support platforms where uint_least16_t != uint16_t.
Out of curiosity, I was trying to track down this claim, but I'm coming up empty. Do you have a source for this?
|
No worries @maspe36, thanks for taking a look.
I'm actually failing to find a direct source for that claim as well, though I thought I had one. I think it can still be inferred from the definitions of |
According to https://design.ros2.org/articles/wide_strings.html, the
stringandwstringtypes must use the UTF-8 and UTF-16 encodings (not necessarily enforced), cannot contain null bytes or words (enforced), and, when bounded, are measured in terms of bytes or words.Moreover, though the rosidl_runtime_c
U16Stringtype usesuint_least16_t, Rust guarantees the existence of a preciseu16type, so we should use that instead ofushort, which isn't guaranteed to be the same asuint_least16_tanyways. (Rust doesn't support platforms whereuint_least16_t != uint16_t.)