[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Another fix (Re: Bugs, bugs, bugs.)
From: |
Just van Rossum |
Subject: |
Another fix (Re: Bugs, bugs, bugs.) |
Date: |
Tue, 1 Aug 2000 14:36:33 +0100 |
At 5:47 PM +0200 31-07-2000, Werner LEMBERG wrote:
>. The attached PS font isn't handled properly.
[ dgjrussian.pfb ]
I've looked at this font, and at FT2's behavior with it:
- It seems that when not autohinting something goes terribly wrong with
scaling the outline; FT_Outline_Get_CBox() return a *huge* bounding box,
hence the failing FT_ALLOC()'s: ftsmooth.c tries to allocate a huge pixmap.
Using non-aliased rendering doesn't seem to make a difference.
[ ...more debugging... ]
Ok, got it, it's that darn FontMatrix parser again ;-). So somehow it *is*
being used. Anyway, the problem was that it didn't handle values of the
form .001 (nor -.001) correctly, and an arbitrary large value was the
result. I've fixed it. (Btw. had atof() been used here, this wouldn't have
been a problem!)
- This is one of the most poorly drawn and produced fonts I have _ever_
seen ;-). Many contours are not closed. The autohinter doesn't choke on
those, but they do give funny artifacts when rendering(*). However, the
open contours doesn't seem to be the big problem: one particular glyph has
one, properly closed, contour, and it still doesn't render without
autohinter.
*) I think it's the t1z glyph loader's fault: it doesn't do the right thing
when there's no closepath before doing a new moveto. I don't see the
correct fix yet, though.
Just
- Another fix (Re: Bugs, bugs, bugs.),
Just van Rossum <=