On 9/1/06, address@hidden <address@hidden
> wrote:Excuse me, I'm confused a bit, please let me get myself together.
I understand as following:
* the reference implementation is WebKit on MacOS X Quartz (only).
* the tested implementation is WebKit on Linux X11.
* the rendering result onto screen by WebKit-on-Linux-X11 is bigger
than that by WebKit-on-OSX-Quartz.
I apologize for having been unclear, but you got the point. You've summarized the problem we have correctly.
If the issue is more generic (e.g. how about on WebKit-on-Win32,
if there's such?), please let me know.
No clue. It is hard enough to bring webkit on linux, we'll think about others platforms in the future ;)
>We tried to hack FreeType to make the two rendering fit. We changed in
>ftobjs.c the values of 72 dpi and the 64 coefficient in
I have 2 questions:
* this is not WebKit specific? for example, comarison between
"text shown by TextEdit.app at specified pointsize" and
"text shown by FreeType2 application on X11/Quartz at same pointsize"
is enough for testing?
Correct. If you port textedit.app using FreeType instead of Quartz, you'll encounter the same issue.
The 72->64 change was a kind of hack just to try to make the rendered text fit the expected metrics... But there is not solid basement on this figure...
Regards,
Jean-Charles