Eli Zaretskii writes: >> From: Mauro Aranda >> Date: Wed, 14 Oct 2020 09:00:27 -0300 >> Cc: 43977@debbugs.gnu.org, schwab@linux-m68k.org, >> Lars Ingebrigtsen , bugs@gnu.support >> >> > See also bug#33566, which is what caused this issue. >> >> Andreas patch looks correct to me. > > I didn't say it wasn't. I didn't mean to imply you said it wasn't correct. Sorry about any confusion. > I just pointed to that bug report, so that > people could (a) see if there were other solutions to consider (e.g., > maybe to solve the original issue in some other way); and (b) so that > the use case in that bug could be retested to make sure this proposed > change doesn't adversely affect it in any way. I think that: (a) The group widget was misdocumented as a possible composite type in the first place: that's why custom-variable-value-create was rejecting it, because it didn't have the format spec it requires. But it was documented, and later the group widget's :format adapted, so maybe it is better to just leave that as it is now. (b) While the change in Bug#33566 is what caused this issue, using "%v" as the format for the group widget is perfectly fine here, and doesn't affect any other uses of the group widget. There is no call to custom-variable-value-create here, so no reason to worry about overriding the :format spec to "%v" at all.