jhalfs: patch for LANG settings + a few fixes

Pierre Labastie pierre.labastie at neuf.fr
Wed Feb 8 10:24:16 PST 2012


Le 08/02/2012 18:41, Pierre Labastie a écrit :
> Hi,
>
> I think I could make the modifications we wrote about
> yesterday. I have tested with:
> lang=fr_FR.UTF-8 full-locale=n
> -> locale -a lists only the minimal set, including
> fr_FR.UTF-8
> lang=POSIX full-locale=n
> -> locale -a lists only the minimal set
> lang=en_GB.UTF-8 full-locale=n
> ->  locale -a lists the minimal set + en_GB.UTF-8
> lang=en_GB.utf8 full-locale=n
> ->  locale -a lists the minimal set
> lang={fr_FR.UTF-8,POSIX,en_GB} full-locale=n
> -> locale -a lists the whole set (and glibc build is 1 mn longer)
>
> My main concern is that I have to rely on LANG being a
> canonical name (i.e. as in the SUPPORTED file). Right
> now, I just ignore it, if it is not found in that file.
> And actually, I do not know what else to do.
> BTW (maybe more appropriate to the LFS list),
> mentioning the `SUPPORTED' file would be of
> some help for beginners (like me a few
> years ago) scratching their heads in front
> of the "7.13 The bash shell startup files" sections on
> canonical names.
>
> Furthermore, the attached patch :
> - adds `check' to the list of packages which are not
>   built when testsuite=0 (same behavior as presently
>   for tcl expect and dejagnu)
> - corrects a typo (fylesystem)
> - suppresses the warnings when compiling conf and mconf
>   (try "make clean ; make" before and after applying the patch)
>   Not a big deal of course, but annoying.
>
> Regards
> Pierre
>
Sorry,
there is a bug:
lang=en_GB.utf8 full-locale=n
->  bombs out.
When I reported the test above, lang was somehow not
updated and the test run with lang=POSIX.

No time now, but a fix'll come later.

Regards,
Pierre




More information about the alfs-discuss mailing list