[poppler] Image optimization

[Albert Astals Cid]

A Dijous 19 Maig 2005 23:14, vàreu escriure:
> On Thu, 19 May 2005 23:02:35 +0200, Albert Astals Cid wrote:
> > A Dijous 19 Maig 2005 22:49, vàreu escriure:
> > > On Thu, May 19, 2005 at 10:47:01PM +0200, Albert Astals Cid wrote:
> > > > Georgeous, can this be applied to the splash outputdev too?
> > >
> > > I think it might be better to leave the splash outputdev as is, only
> > > changing it is as necessary. This lets us keep a stable reference and
> > > encourages people to fix/use the cairo backend.
>
> I'm a little confused. If the fix is not related to cairo, yet the fix
> helps only the cairo backend, then that suggests there is an
> abstraction problem somewhere in poppler, (eg. two copies of the same
> code where there should be one).

I'd say we have such a problem, SplashOutputDev and CairoOutputDev have 
roughtly 50% code that is the same.

> Now, if there are backend-specific fixes that help one poppler backend
> outpace another, then there's not much that can be done to prevent
> that.
>
> > I disagree, keep for sure that if ever i get myself to use poppler for
> > future kpdf versions i'm not going to ask kde people to install cairo if
> > it is not absolutely necessary. And for now i don't see any reason more
> > than "we are uber cool and use the buzzworder cairo". No offense to cairo
> > developers, it's a cool technology but it's a bit much to ask people to
> > install cairo instead of using splash when both have a similar
> > performance.
>
> I'm certainly not offended. I wouldn't ever encourage someone to
> switch to cairo unless there were a quality or performance benefit
> they could benefit from, or if there's some other direct cairo
> benefit, (eg. instant support for multiple backends or something).
>
> And I know that in many situations, cairo just isn't there yet on the
> performance front, (but we're planning to put a lot of effort into
> cairo performance particularly over the next month or two).
>
> >                                                       Optimizing only
> > cairo and not splash is just another way of beign more gnomeish than
> > kdeish.
>
> I wouldn't argue for restricting a fix to the cairo backend if it
> applied just as well to the splash backend, (see comment above on the
> broken abstraction).
>
> But I do have to say that I don't understand your reference to gnome
> and kde in this context. Has something made made you think that cairo
> is more directly associated with gnome than kde?
Well, it's the general feeling everyone has, maybe because most of the devels 
seem to be more gnome devels than kde devels, also maybe because Qt4 is 
supposed to have Arthur that is similar to Cairo. 

> And what can I do to help change that? 
I think you can do little about it other than sending mails like this each 
time someone mentions Cairo is more gnome related than kde related.

Albert

> That's an impression I certainly don't want anyone 
> to be left with. I'd like cairo to be usable by any desktop
> environment.
>
> -Carl