On Wed, 26 Jan 2011 13:19:07 +0100 Dieter Plaetinck <dieter@plaetinck.be> wrote:
On Mon, 24 Jan 2011 17:22:18 -0500 Matthew Gyurgyik <pyther@pyther.net> wrote:
Falconindy was right. nullglob affects everything just not the function. I simply fixed it by adding shopt -u nullglob at the end of the function.
I rebased and forced a push therefore there is no "new" commit: https://github.com/pyther/aif/commit/43d02dc58a63599ff67d3a7e9f89c8b81831b56...
I don't think it is worth it to resend the patch yet again for one line of code, though let me know if that is what you want me to do.
can you be sure that nullglob must be unset after the function? Could it be that nullglob was already set in the beginning?
I can answer my own question: http://mywiki.wooledge.org/NullGlob "nullglob is not on by default because there are other cases where its behavior would be extremely disconcerting. For example, ls behaves quite unexpectedly if an unmatched glob is removed from the argument list:" I actually found some occurrences in the aif codebase where I enable nullglob, probably the right way forward is to disable those again after the "critical codepath", check if i didn't break anything else, and then merge your patch. Dieter