bug#58985: 29.0.50; Have auth-source-pass behave more like other back ends

[J.P.]

Hi Kai,

Kai Tetzlaff <emacs+bug@tetzco.de> writes:

> This change breaks my use of `auth-source-pass' in gnus.

Thanks a lot for reporting this. And sorry about the breakage.

> I haven't had time to investigate the issue but what I can already say
> is that the problem occurs independent of the value of
> `auth-source-pass-extra-query-keywords' (`t' or `nil'). So the
> change is not backward compatible. It would (at least) be nice to
> mention this in the NEWS entry.

I'd rather not settle for "at least" if we can help it. If the user
option doesn't preserve existing behavior, that's a bug that needs
fixing.

The traditional and new code paths diverge in `auth-source-pass-search',
so without a backtrace, we should start there. (Obviously, a full
backtrace would be ideal, but I understand completely if you're not
willing to surrender one.) First off, can you try reverting the changes
to that function alone? Just eval'ing a modified version in place,
without the extra `cond' clause and the two keywords, :max and :require,
should do it. If that doesn't tell us anything (and only if you're up
for it) you could trace the function and tell me what the inputs were
(obviously after swapping out any sensitive info). A mini example of
your ~/.password-store layout might also be helpful.

According to etc/AUTHORS, you're likely much better acquainted with
Emacs than I (2009!). So, please adjust the above recommendations
accordingly and, if possible, apply some of that experience to helping
fix this bug. And apologies again for the disruption.

Thanks,
J.P.