Re^4: Memory Leak with XS but not pure C

uc suffers from The Unicode Bug when the unicode_strings feature isn't in enabled.

It works correctly (giving SS for ß) when the string the unicode_strings feature is enabled.

It works correctly (giving SS for ß) when the string is stored in the UTF8=1 format.

It works incorrectly (ß unchanged) otherwise.

use open ":std", ":locale";
use feature qw( say );

my $ss = "\xDF";
utf8::upgrade(   my $ss_u = $ss );
utf8::downgrade( my $ss_d = $ss );

{
   no feature qw( unicode_strings );
   say uc( $ss_d );                    # ß
   say uc( $ss_u );                    # SS
}

{
   use feature qw( unicode_strings );
   say uc( $ss_d );                    # SS
   say uc( $ss_u );                    # SS
}
[download]

Comment on Re^4: Memory Leak with XS but not pure C Select or Download Code

Replies are listed 'Best First'.
Re^5: Memory Leak with XS but not pure C by FrankFooty (Novice) on Mar 31, 2025 at 10:21 UTC
Thanks very much Ikegami. I hadn't heard of that feature How is it possible to get the same behaviour within an XS function	[reply]
Re^6: Memory Leak with XS but not pure C by syphilis (Archbishop) on Mar 31, 2025 at 12:29 UTC
How is it possible to get the same behaviour within an XS function There is https://perldoc.perl.org/perlapi#Character-case-changing . (Maybe other places, too ... but that's the only link I could find quickly.) Cheers, Rob	[reply]
Re^6: Memory Leak with XS but not pure C by ikegami (Patriarch) on Mar 31, 2025 at 13:20 UTC
If you're asking how to convert to uppercase from XS, you'd start with an upgraded string, then call `toUPPER_uvchr` or `toUPPER_utf8` repeatedly. (Or call a Perl function that uses `uc`.)	[reply] [d/l] [select]
Re^6: Memory Leak with XS but not pure C by ikegami (Patriarch) on Mar 31, 2025 at 13:18 UTC
There is nothing that can affect the behaviour of Perl's `uc` operator in XS code. But you can't use Perl operators in C/XS code anyway.	[reply] [d/l]