[perl #133782] length($^R) sometimes wrong

On Sat, 19 Jan 2019 05:45:44 -0800, haukex@zero-g.net wrote:
> Hi everyone,
> 
> This oneliner fails, when it most likely shouldn't:
> 
>   perl -e 'while( "\N{U+E4}bc" =~ /(..?)(?{$^N})/g )
>            { length($^R)==length("$^R") or die }'
> 
> A slightly longer test program (attached) shows that on the second 
> iteration, length($^R) is reporting 2 instead of the expected 1. It 
> seems to have something to do with UTF8 strings, since it works fine on 
> non-UTF8 strings.
> 
> Tested on several versions of Perl on Linux and Windows (down to 5.8.9).

From a debugging build:

 $ ./perl -e 'while( "\N{U+E4}bc" =~ /(..?)(?{$^N})/g ){ length($^R)==length("$^R") or die }'
 panic: sv_len_utf8 cache 2 real 1 for c at -e line 1.

so I suspect a missing SvSETMAGIC() somewhere.

Tony

---
via perlbug:  queue: perl5 status: open
https://rt.perl.org/Ticket/Display.html?id=133782

• [perl #133782] length($^R) sometimes wrong by Tony Cook via RT
• [perl #133782] length($^R) sometimes wrong by Tony Cook via RT
• Re: [perl #133782] length($^R) sometimes wrong by Hauke D
• [perl #133782] length($^R) sometimes wrong by James E Keenan via RT
• Re: [perl #133782] length($^R) sometimes wrong by Hauke D
• [perl #133782] length($^R) sometimes wrong by James E Keenan via RT
• [perl #133782] length($^R) sometimes wrong by Hauke D