[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-core-devel
Subject:    Re: Making kdefx static
From:       Matthew Woehlke <mw_triad () users ! sourceforge ! net>
Date:       2007-08-03 22:48:43
Message-ID: f90bcc$m7n$1 () sea ! gmane ! org
[Download RAW message or body]

Allen Winter wrote:
> I like the idea of having Blitz in kdesupport.
> But... only if it is "done" or "nearly done".

Mosfet wanted me to post this:
<quote src="Mosfet">
Hey guys, I'm sitting at home reading this thread wondering why every 
time I'm involved in something it turns so complex >:)

For people worried about stability since it is new code I'm doing *a 
lot* of testing. More than I ever did before on anything I've written. 
Each effect is being tested individually with an 8bit, 32bitRGB(A), and 
premultiplied image. I spent all morning doing this. Next I am going to 
test with MMX disabled so the C/C++ versions are all tested.

KImageEffect OTOH really needs to go. Someone just ran the porting 
script over it and it does all sorts of stupid things like treating 
QImage::colorTable() like it's a reference and not a deep copy. It's all 
sorts of new bugs over old bugs.

As for people suggesting developers just using their own copies of 
KImageEffect, two problems with that: First of all, like I said most 
stuff in KImageEffect doesn't work anymore. Apps using their own copy 
doesn't fix that. Second of all there are also local copies of 
post-KImageEffect algorithms I've done scattered all over the place. 
Blitz fixes and integrates all this.

So anyways, my vote now is for kdesupport.
</quote>

> And, keep in mind that it will need to follow
> the licensing and other requirements (Krazy, BIC).

Obviously Mosfet would need to speak to that. (Can someone please 
un-moderate him so he can contribute reasonably to these discussions?)

-- 
Matthew
"Non sequitor. Your facts are out of order." -- Nomad

[prev in list] [next in list] [prev in thread] [next in thread]