Loud ramblings of a Software Artisan

Wednesday 4 May 2005

Apple developers clarifies for KHTML

I talked recently about KHTML and Apple issues, where KHTML express their concern about reintegrating Apple change to KHTML and how uneasy it was.

Dave Hyatt, feeling much obliged, replied to restart the dialog. KHTML clarified what they wanted. I personally doubt they ever get the bug database shared, as even inside Apple R&D staff the bug database access is restricted. I remember to have entered a bug about some project, I couldn't even read it after because I had no privileges. Some other seems to be doable, but for the same reason, I don't see Apple committing directly to KHTML source control, and I don't see Apple sharing the CVS of their KHTML directly for the same practical confidentiality implications.

Wait and see...

Chateau de Castelnau - Picture of the Day, May 4th, 2005

I haven't posted any picture for some time. So here is one from the film archives, those that actually may stay longer than the proprietary RAW files the digital camera manufacturer want us to use....

This is Chateau de Castelnau ; it is one of the last picture of tour of Perigord, France, in September 2001.

Provia 100F slide film pushed 2 stops (hence the quite high contrast), still with my Canon Elan 7/EOS 33 using a EF 24-85 f/3.5-4.5 USM lens. 1/90sec f/32.

Major AbiWord MacOS X fixes

AbiWord on MacOS X had broken justified text display for a few month, and that is really bad. See bug 8384 and all its duplicates. I hadn't used AbiWord on Mac for some time, and I wanted to update some document I had on a CD that the crappry CD-ROM drive in my PC is unable to read. So I did read it on the MacMini and used AbiWord for that. That is what made me fix that bug. Not that I wouldn't have, but I was not really aware of that as it was working when I left the code last year.

It appears that a change in December to support Bi-directional languages removed the code to draw word by word justified text and use the inter-word gap as specified to the GR_Graphics::drawChars(). The UNIX code use Xft to draw chars one by one, but MacOS X is not very fast to do that.

So I reimplemented that with the new code, but broke the bi-directional languages support (programming by side effects). That will be fixed differently as since Tomas has added hooks to tell whether we support that or not. At the same time I commented that part to explain why I do that. I expect to have simplier code by then. Ultimately I may rewrite this using ATSUI.