- From: Ian Hickson <ian@hixie.ch>
- Date: Sun, 15 Sep 2002 14:49:27 +0000 (GMT)
- To: Henri Sivonen <hsivonen@niksula.hut.fi>
- Cc: "www-style@w3.org" <www-style@w3.org>
On Sun, 15 Sep 2002, Henri Sivonen wrote: > > It would likely be bad for the performance of the (hopefully upcoming) > fixes for ATSUI-related Mozilla bugs (such as 121540 and 165878) if > Mozilla wouldn't be allowed to just pass a list of fonts in the > preferred fallback order (based on author CSS and user prefs) to ATSUI > and had to implement different fallback rules on the application side. I want UAs to try all fonts on the system for each glyph, not just the specified ones. Also, the specified fonts have to be selected based on the rules in CSS1 (which decide which fonts 'match', e.g. bitmapped fonts and sizes, italic, etc), and which font is appropriate for each glyph has to be based on the unicode-range descriptor as well as other factors. So tough, you can't just let the OS do it. You can use the OS for some aspects, but not all. > So far, Mozilla's attempts to avoid ATSUI and to implement everything > on the application side have lead to worse text rendering results than > OmniWeb's approach to go with the OS services. No offence to the Omniweb people, but it's hardly an example of great CSS compliance. Sure, if you ignore the spec, you can make your implementation blazingly fast. Just skip any graphics support and dump the markup to the screen. And no offence to the Mozilla people, but it's hardly an example of optimal engineering either. So neither of these UAs are good examples. -- Ian Hickson )\._.,--....,'``. fL "meow" /, _.. \ _\ ;`._ ,. http://index.hixie.ch/ `._.-(,_..'--(,_..'`-.;.'
Received on Sunday, 15 September 2002 10:49:29 UTC