Hi again!
> The LC_COLLATE=C option is tested and works as expected in sh.
Justin Pryzby was still right and I was wrong. I tried again, and it needs to be
LC_ALL=C as Justin said.
Sorry for multiple e-mails. This should be the last version of the patch from my
part.
--
Juhan Ernits
117c117,1
Hi!
I think there are two options:
a) apply the attached revised patch (2 lines) to /sbin/MAKEDEV. This allows some
calls (for example chown, mkdir etc) to return error messages in locale specific
languages, while the potential pitfalls due to collating sequence are taken care
of.
b) add
expor
You can just export LC_ALL=C at the top of the file, then you can
assume sorting order and other stuff. If you just set LC_COLLATE or
similar, then the users value of LC_ALL will override it (if valid,
otherwise it will fall back to C anyway).
Justin
--
To UNSUBSCRIBE, email to [EMAIL PROTECTE
Actually it should be there only for the locale dependent invocations, thus only
for the sed invocation on line 117 in /sbin/MAKEDEV.
But, having looked at the script closer, it looks like LC_COLLATE=C
should be additionally set globally as there are cases (for the case clause
starting on line 32
On Mon, 2006-04-17 at 10:05 +0300, Juhan Ernits wrote:
> $ echo pqrstu | LC_ALL=et_EE sed -e 's/[a-z]//g'
Is the LC_ALL=C necessary for all the sed invocations, or only the ones
doing an explicit alpha range match? If it's necessary for all calls,
is it possible to set this once for the script
Package: makedev
Version: 2.3.1-81
Severity: normal
Tags: patch l10n
*** Please type your report below this line ***
As Denis Barbier pointed out, z is not the last character of the
alphabet in some locales:
$ echo pqrstu | LC_ALL=et_EE sed -e 's/[a-z]//g'
tu
Thus sed should be called un
6 matches
Mail list logo
