RE: [21.4,21.5] Metacity maximize bug localized
There have been persistent reports of inflooping with window
managers that don't respect XEmacs's desire to size to a
exact number of character cells, specifically when maximizing
the XEmacs frame. The surface problem is that XEmacs doesn't
take XtGeometryNo for an answer. The attached patch fixes
that, and my XEmacs is now well-behaved under metacity.
However, the underlying problem is that the EmacsManager
shell widget's geometry management system contains mutually
recursive functions, and furthermore actually changes the
geometry of children in the process of a query. Really
straightening this out (at minimum, the structure of the
geometry manager must be documented) isn't going to be easy
(unless there's somebody more expert than I willing to work on it).
Well, I do have a workspace with a bunch of sizing fixes in it; but I'm not
sure whether it addresses this problem.
Btw the whole point of the EmacsManager is to be as simple as possible and
just let XEmacs do the controlling; at the time I wrote this, it blew my
mind that there was no simple widget to do this. We had tried using nested
Pane widgets under Athena, e.g., but it just didn't work.
Apply this patch to 21.5 and work out any new geometry
management issues as they come up, then get something into
21.4 quickly.
I say, go ahead.