Re: Another question on the W32 port

[Kim F. Storm]

Jason Rumney <address@hidden> writes:

> address@hidden (Kim F. Storm) writes:
> 
> > I was looking at glyph_rect in X and W32, and noticed that they are
> > significantly different due to the following change:
> > 
> > 2002-02-18  Jason Rumney  <address@hidden>
> > 
> >     * w32term.c (glyph_rect): Determine the row and glyph more precisely.
> > 
> > 
> > Now I wonder why that change was not back-ported to the X version?
> > 
> > Was there a specific W32 problem that was related to this, or can the
> > new W32 version be used on X too (modulo the differences in rectangle
> > implementation) ?
> 
> My memory that far back is sketchy, but I think it may have been to
> do with mouse highlighting. There have long been bugs in mouse
> highlighting that show up under Windows, but not X, and it has often
> confused me when I have found a fix for such bugs and wondered why the
> fix was not necessary on X. 

Yes, that is strange.

Maybe it can be related to cleartype as you mention in the below
W32-specific comment from get_glyph_string_clip_rect:

  /* If drawing the cursor, don't let glyph draw outside its
     advertised boundaries. Cleartype does this under some circumstances.  */


>                             In such cases I decide to play it safe
> and not backport the change to X.

I would prefer if such "improved" code was backported (or
consolidated) like I did with get_glyph_string_clip_rect, i.e. let the
additional checks be conditioned by HAVE_NTGUI.  That way, if similar
problems are seen on other platforms later on, the fix may actually be
there already.

-- 
Kim F. Storm <address@hidden> http://www.cua.dk