On Wed, Oct 31, 2007 at 08:01:26PM +0200, Timo Korvola wrote: > Luis Rodrigo Gallardo Cruz <rodrigo nul-unu com> writes: > > The idea itself ought to be uncontroversial. > > Actually, I am not entirely convinced of the usefulness of these > properties. Are there characters in UTF-8 that cannot be > represented as COMPOUND_TEXT? There most likely are. I don't really know anything about writing systems, but the spec for COMPOUND_TEXT does not mention, for example, any of the languages of India, but it does have some sort of extensibility mechanism. I have no idea how likely is a random xlib to have support for that. > > In order to get everything working I made sawfish assume > > all strings to be rendered to the screen are encoded as UTF-8. > > That is useful regardless of the _NET_WM properties, because the host > encoding Sawfish happens to be using may not be able to represent all > necessary characters even if COMPOUND_TEXT is. Unfortunately this > change is clumsy to implement and without benefit if > Xutf8TextPropertyToTextList is not available. With > XmbTextPropertyToTextList conversion has to go via the local encoding > anyway. Which is another part of what I wanted to bring up. How many people are still using systems without Xutf8Text*? How much would we break if we started assuming we have all those modern thingies such as Gkt 2? > > Anyways, I'm requesting feedback on the patch. > > What version is it against? It did not apply cleanly to the trunk. Ups. Sorry about that. It's made against 1.3.1+<debian patches> > Anyway, I applied the rejects by hand, fixed some errors and just > cleaned it up a bit. This patch is against the current trunk (r4204). > Alternatively you can pull branch utf-8-names from > http://www.iki.fi/tkorvola/sawfish.git. Are you mirroring svn there? I'll probably clone that for my own use if so. -- Rodrigo Gallardo GPG-Fingerprint: 7C81 E60C 442E 8FBC D975 2F49 0199 8318 ADC9 BC28
Attachment:
signature.asc
Description: Digital signature