[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [AUCTeX-devel] GNU AUCTeX branch, master, updated. fa1daf8cf4c624dc1
From: |
Tassilo Horn |
Subject: |
Re: [AUCTeX-devel] GNU AUCTeX branch, master, updated. fa1daf8cf4c624dc17f584d821f774527579cc97 |
Date: |
Wed, 12 Mar 2014 09:00:09 +0100 |
User-agent: |
Gnus/5.13001 (Ma Gnus v0.10) Emacs/24.3.50 (gnu/linux) |
Mosè Giordano <address@hidden> writes:
>> Why this change? My `completing-read-multiple' never returns nil, if
>> no input is inserted it returns `("")' which isn't nil. Am I missing
>> something? I'm running GNU Emacs 24.3, if that helps.
>
> Ok, the behavior of `completing-read-multiple' has been changed since
> the release of GNU Emacs 24.3 with this commit:
> http://bzr.savannah.gnu.org/lh/emacs/trunk/revision/112063
And I wondered how you didn't notice that bug when testing. ;-)
> The return value of `completing-read-multiple' with empty input is
> used elsewhere (e.g. in style/biblatex.el), so we have, at least, two
> possibilities:
> - everywhere check if the return value is `nil' or `("")';
> - treat the two different behaviors inside
> `TeX-completing-read-multiple' and always use it.
Yes, making TeX-completing-read-multiple always return nil on empty
input is definitively the best choice, so I've just implemented that
including the same behavior for the XEmacs `multi-prompt' version.
Also, now all code calls `TeX-completing-read-multiple' instead of just
`completing-read-multiple' (in biblatex.el), and I've adapted hopefully
all checks for empty input.
Bye,
Tassilo