Re: emacs-29 00de296d1b 3/3: Simplify erc-sasl's auth-source API

[F. Jason Park <jp@neverwas.me>]

Po Lu <luangruo@yahoo.com> writes:

> Amin Bandali <bandali@gnu.org> writes:
>
>> branch: emacs-29
>> commit 00de296d1b4f629fd828cdeff588bb4f742d9ffe
>> Author: F. Jason Park <jp@neverwas.me>
>> Commit: Amin Bandali <bandali@gnu.org>
>>
>>     Simplify erc-sasl's auth-source API
>>     
>>     * doc/misc/erc.texi: Revise descriptions in SASL chapter to reflect
>>     simplified auth-source options.
>>     
>>     * lisp/erc/erc-sasl.el (erc-sasl-password,
>>     erc-sasl-auth-source-function): Revise doc strings.
>>     (erc-sasl-auth-source-password-as-host): New function to serve as
>>     more useful choice for option `erc-sasl-auth-source-function'.
>>     (erc-sasl--read-password): Promote auth-source to pole position, above
>>     an explicit string and `:password'.
>>     
>>     * test/lisp/erc/erc-sasl-tests.el (erc-sasl--read-password--basic):
>>     Massage tests to conform to simplified `erc-sasl-password'
>>     API.  (Bug#29108.)
>
> Was such a big change really a regression fix, suitable for emacs-29?

My apologies. Won't happen again.

The file it updated was first introduced five days back and contained a
poorly designed option (by me) that threatened to become a maintenance
nuisance and a possible source of user frustrations (and maybe bug
reports). Basically, it suggested a new type (symbols) be assigned to a
decades-old variable, `erc-session-password', without a thorough vetting
of the repercussions. The change in question tried to undo some of that
short-sightedness, albeit too elaborately (and thus irresponsibly),
given the timetable.

(FWIW, that should be the last you'll hear from ERC for Emacs 29, other
than incrementing our version number to 5.5 to keep the releases in
sync.)

Appreciate your vigilance and understanding (and all your work on
Emacs).