[pve-devel] [PATCH v3 storage 2/3] Incorporate wipe_disks from PVE::Ceph::Tools

[Thomas Lamprecht]

On 4/16/20 7:41 AM, Dominik Csapak wrote:
> wanted to chime in here (since i believe the original comment
> about unicode literals came from me)
> 
> On 4/15/20 5:02 PM, Thomas Lamprecht wrote:
>> albeit I'm not to sure if the unicode stuff is really true, I tried:
>> perl -we '"Ⅲ" =~ /\d/ or die "no match"'
>>
>> where "Ⅲ" is the roman numeral three (U+2162) and that does not matches,
>> but it could be wrong encoding or whatever...
> 
> if you use utf symbols in the source, it only works if you
> have 'use utf8;'
> 
> also the roman numeral three is not considered (afaik) as
> part of the 'digits' group (only those that also match \p{Digit} [0])
> but '৪' for example (bengali digit four (U+09EA))
> 
> if the source comes from outside, you have to have a 'perl string'
> with the correct code point,
> which happens e.g. when using decode_utf8 on a string with
> such a character
> 
> so this example matches:
> 
> ---8<---
> use Encode;
> my $val = decode_utf8(shift);
> if ($val =~ /(\d+/)) {
>     print "found $1\n";
> }
> --->8----
> 
> when calling with
> perl foo.pl ৪
> 
> so to avoid confusion (and probably errors), i proposed always using [0-9] instead of \d
> 
> 0: https://perldoc.perl.org/perlrecharclass.html


Thanks for clearing that up, I mostly poked at it as Dominic came to me
asking what to do as you told him to use [0-9] and Fabian told him to go
back again, not knowing about your suggestion.

The question is, does the matching of other digit really matters at all?

I mean it naturally depends on what happens whit the parsed data, but most
of the time it should not really matter at all, or?