bug#61496: 30.0.50; Default value of icon-title-format

["Óscar Fuentes" <ofv@wanadoo.es>]

Eli Zaretskii <eliz@gnu.org> writes:

>> >   . I suggest that the "special" value which means "use
>> >     frame-title-format" will be t, not nil, so that users who want it
>> >     set it explicitly, not by some omission.
>> 
>> The purpose of changing the default is based partly on the idea that,
>> very likely, the user will want to keep icon-title-format in sync with
>> frame-title-format. So not having to do anything to achieve that is a
>> feature, not a bug.
>
> We are mis-communicating.  My point is that the value nil means "don't
> display this".  E.g., try setting mode-line-format or
> frame-title-format to nil, and observe the effect.

I don't object to use `t' instead of `nil'. I object to leaving the
default as is.

>> > and it is quite possible that someone out there does want/expect
>> > the default value to be a string.
>> 
>> I have no idea how this can be.
>
> Please believe me and my gray hair: with Emacs, anything that is just
> possible is usually done by someone out there.

You already have a real case of a broken setup (mine). Furthermore, the
people that depend on the contents of the window title to automate
things on their machines are the most likely to set frame-title-format,
and thus it is practically guaranteed that having a working
icon-title-format on its current incarnation will be problematic for
them.

>> If this is your final decision, it's better to discard the whole
>> proposal, as it gives little benefit.
>
> I disagree.  I think it will allow you and others in your situation to
> have the problem solved,

icon-title-format already broke my customization, and I already fixed it
on my .emacs.el. This is not about me, it is about avoiding the pain I
went through to others.

> while not risking breaking someone's
> unrelated customizations.  I don't understand the urge to have it
> solved automagically, even for those who are affected by the changes
> in Emacs 29.

This is about a real group of users for which this issue can cause lots
of disruption and frustration (Emacs' window title switching to the
contents of icon-title-format when the user jumps to other virtual
desktop is far from obvious) vs an hypothetical group of users that
somehow put on their config an effectless variable on a way that causes
the interpreter to throw an error on their face.

For which group do you expect more bug reports?