Hi,

On Mon, Jul 15, 2002 at 01:53:02 -0700, Sean 'Shaleh' Perry wrote:
> On 15-Jul-2002 Mike wrote:
> > Hey,
> > 
> > I updated Johannes Winkelmann's 0.65.0alpha2 aa patch, originally from
> > http://www.hta-bi.bfh.ch/~winkj/. Works pretty well :)
hehe, nice... actually I have a patch for beta2 as well, but as there
was very little feedback/interest I didn't upload them. And there is a
drawing error in the toolbar/clock code [1] and didn't have the time to
look into it (it's basically just translating
  (font->max_bounds.rbearing - font->min_bounds.lbearing) * 2
to Xft)).

> AA support will be interesting.  I know people want it but it is XFree86 4.x
> specific.  Which means more ifdefs.  Brad also believes AA is a crock and a
> waste of time.  Personally I have not seen much benefit either way.
>
> This is something I plan to look at after 0.65.0 is released.  xOr has a nifty
> approach we may consider using.
Not sure what xOr's approach is, but I was wondering whether you could
plan to abstract the fonts into a separate class. If the AA patch would
work for multibyte as well, we would have code like

if ( isMultiByte ) {
  if ( useXft ) {
    ...
  } else {
    ...
  }
} else {
  if ( useXft ) {
    ...
  } else {
    ...
  }
}
    
for both getting the text width and drawing it. This kind of code is
in many places (in the original source without the xft blocks of
course), at least that's my impression from writing AA-patches for
0.62.1, 0.65.0{alpha{6,7,8},beta{1,2}} :-)

If this could be wrapper into a separate class, the #ifdef would be in
one place only and therefore not really disturbing the source (except
for the font code). Plus it would simplify the rest of the code a bit
by remove some duplicated code - even if you don't plan to implement AA 
support for now.

Just a thought...
Regards, Johannes

References:
  1 http://www.hta-bi.bfh.ch/~winkj/files/clock_aa.jpg
-- 
Johannes Winkelmann

Reply via email to