Hey Pierre. On Sun, 2023-05-21 at 14:23 +0200, Pierre Gruet wrote: > Could you please tell me if the issue is still showing up in the > currently packaged version, and if so, can you please give me the > steps > to reproduce it?
I've just tried and the outcome is a bit odd: While I don't get the error anymore, with set -u in place, the only use of the variable: https://salsa.debian.org/science-team/scilab/-/blob/09ef844d98df2d4967a465ecd9d175c39893948b/debian/bash_completion.d/scilab#L52 should in principle still fail if it's not set (i.e. it uses just $filename and not something like ${filename-} ... which would initialise it to empty if unset. That would mean that something else from bash_completion already sets it now. But, if I add the following in the completion file: set >> /tmp/foobar before the complete -F _scilab $filenames scilab scilab-cli scilab-adv-cli and then cause some completion (otherwise it won't execute the file)... /tmp/foobar does not show that filenames would be set... Even more weird... in principle it should still fail. But apart from that... I don't even see the reason why the variable is there at all (in the completion). Because AFAIU in: complete -F _scilab $filenames scilab scilab-cli scilab-adv-cli the "$filenames scilab scilab-cli scilab-adv-cli" are the "name..." argument(s) to complete and -F _scilab is the function executed when any of the name(s) is completed. Then it wouldn't make sense to have $filenames in the list. Maybe -o filenames (no $) was meant? Cheers, Chris. -- debian-science-maintainers mailing list [email protected] https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/debian-science-maintainers
