Re: nocaseglob

[Bruce Korb]

On 1/22/07, Bob Proulx <bob@proulx.com> wrote:
> Unfortunately in this case setting the variable on the command line
> for that command won't do what you want.  Since file globbing takes
I figured that out.  :}

>   (unset LC_ALL LC_COLLATE LANG ; echo [a-z]*)
>   (LC_COLLATE=en_US.UTF-8 ; echo [a-z]*)
>   (LC_COLLATE=C ; echo [a-z]*)

$ echo [a-z]*
bin bk-archives Bugzilla core cron Desktop docs [[...]]
$ export LC_COLLATE=C
/home/usr/bkorb
$ echo [a-z]*
bin bk-archives core cron docs [[...]]
$ unset LC_COLLATE
$ echo [a-z]*
bin bk-archives Bugzilla core cron Desktop docs [[...]]
$ echo [a-d]*
bin bk-archives Bugzilla core cron docs
$ echo [a-D]*
bin bk-archives Bugzilla core cron Desktop docs Downloads

OK.  I'm just going to set LC_COLLATE to "C" and not sweat it.
There seems to be something else aside from LC_ALL, LC_COLLATE
and LANG that affects collating sequence.  Notice the difference
between "[a-d]*" and "[a-D]*".  Whatever collating sequence is being used
do select files puts "D" after "d", though the actual sort puts them as being
equal.  Whatever.  I've gotten bored now. :)

> I do not see that behavior here.  I am using the stock bash
> distributed by Debian.

I get the behavior with the stock RedHat and (likely, though not certain)
stock Debian BASHes.  So, it seems clear: something aside from LANG,
LC_ALL and LC_COLLATE affects the collating sequence.

$ unset LANG LC_COLLATE LC_ALL
$ echo [a-D]*
bin bk-archives Bugzilla core cron Desktop docs Downloads

> You might be chasing a patch bug.

Whatever it is, I have the workaround:  LC_COLLATE=C.
Thank you, Bob.

Regards, Bruce