On 03/01/2012 07:44 PM, Jérémy Compostella wrote:
> Hi Pádraig,
>
> Finally, I found time to work again on this patch. I provided it as
> attachment.
>
> I'm not completely satisfied with the documentation part. I've tried to
> be more specific but it becomes quickly complicated. So I get back the
> original explanation which does not completely satisfied me since it
> does not explain the following point:
> - When -a is not specified:
> - output file names are considered exhausted when the first suffix
> character will become 'z'.
> -'z[a-z]' suffixes are never used
> - the fact that suffix length is increased each time the suffix are
> "exhausted".
> But maybe it's not relevant to be so specific.
>
> What is your opinion about this point ?
>
>> Good. That's what I'd prefer anyway so as to be compatible with old
>> data sets. Note '.' sorts before digits (-d) too, so there should be
>> no ordering issues with --additional-suffix=... either.
> In fact, it depends on the current locale. With my locale, "." is not
> sorted this way.
That's problematic.
Is your locale ar_SA? as that's the only one
on my system that triggers:
for loc in $(locale -a | grep -vF .); do
echo $loc; printf "%s\n" 2 1 . | LC_ALL=$loc sort
done | grep 1 -B1 | grep _
Note the ar_SA case may be just a bug in glibc's LC_COLLATE
definition for that locale?
I'll not get to review your patch tonight.
Probably tomorrow.
cheers,
Pádraig.
