bug#74339: 30.0.92; CC Mode stomps C TS Mode

[Stefan Monnier]

>> Indeed, I think loading a package does not show a strong user
>> preference.  As we've seen, it can happen "by accident" in various
>> circumstances that are only loosely related to using the code in
>> that file.
>> Invoking a mode is a significantly stronger signal.

Dmitry wrote:
> Stronger yes, but is it strong? One might as well be testing the mode,
> without intention to set it up as the default.  And how much does it
> help anyway?  The result can only last until the end of the current
> session, after which the user would need to use one of the
> other approaches.

I can't disagree.

Eli wrote:
> But AFAIU the code in both c-ts-mode and cc-mode is written such that
> just invoking the mode (after the mode's package was loaded once)
> doesn't modify major-mode-remap-defaults.

Indeed.  I had proposed a patch which sets `major-mode-remap-*` when we
enable the mode rather than when we load it, but it was rejected because
supposedly it did not correspond to the "least bad" compromise that
resulted from a long discussion.

> So unless I'm missing something, the current code doesn't support what
> you think should be the way for users to express their preference.

AFAIC, "the way for users to express their preference" is by customizing
`major-mode-remap-alist` and is a different issue than the setting of
`major-mode-remap-default` we're discussing, which is about trying to
*guess* the user's preference.


        Stefan