Front page | perl.perl5.porters |
Postings from July 2013
Most if not all the POSIX functions that take a string scalar as input
treat that scalar as bytes, ignoring the UTF8 flag. This is clearly
wrong. But what to do?
One possibility raised on irc is to downgrade to bytes first. If the
downgrade fails, return FALSE and raise a 'wide character' warning.
Another possibility is to instead call the libc wide character
equivalent function using the code point given by the UTF-8 character.
I believe this is more in keeping with DWIM.
For example, POSIX::isalnum("\x{100}") could fail and warn, or it could
return the result of libc iswalnum(0x100).
If we did the latter, Configure would have to be changed to probe for
these functions. I think if one is present all the others of the given
class also would be. If not present, "failing and warning" would be done.
Thread Next