17 Sep
2004
17 Sep
'04
10:38 a.m.
We can test for the presence of Marlett and fall back to the existing code if we don't find it.
That works but then we're maintaining two codepaths, one or other of which will hardly be tested at all. Why not just check in the font binaries? I know CVS isn't very good at it, but it does work and it'll probably save pain later.