bug#22105: 25.1.50; REGRESSION - `modify-frame-parameters'

[Drew Adams <drew.adams@oracle.com>]

> >> Drew, did this fix your problems here?
> >
> > No, this is still broken - since October 2015.  I have tried
> > a few times to track this down (spent quite a while on it),
> > but I have not had a block of time to do it successfully.
> >
> > It is also the case that the behavior, although still broken,
> > has changed several times since then - quite different
> > breakages, due to changes in the Emacs code (I have not
> > changed my code).
> >
> > That is another reason that I have not found the time or
> > energy to track this down completely: it kept changing,
> > as other users reported problems that were similar or
> > (I assume) related.
> >
> > I still hope to get to the bottom of this.  And it seems
> > that the changes of the Emacs code have quieted down now.
> 
> That was 3 years ago, so this is just a friendly reminder.
> Did you ever get a chance to track this down?

No, I haven't had the time.

I just went through the recipe again, and I see the problem.
It's 100% reproducible (I'm using Windows 10).

It would be great if someone knowledgeable about the C
code (e.g. `modify-frame-parameters') would follow the
recipe on Windows and track down what needs to be done.

The bug report identifies approximately when the
regression was introduced.  It would really be great
to have this fixed, restoring pre-regression behavior.