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

List:       koffice-devel
Subject:    Re: Solution for absurd filter chains
From:       Nicolas Goutte <nicolasg () snafu ! de>
Date:       2004-10-03 17:25:25
Message-ID: 200410031925.26756.nicolasg () snafu ! de
[Download RAW message or body]

I know now why I have never tried to use weights to solve that problem.

Weight are not used before displaying the mimetype names to the users, they 
are only used later to find the needed filters (the best way with the minimum 
of weight) but that is too late.

So we are somehow back at start of the problem with 2 ways of finding a 
solution:
- make a quick hack
- change the current filter system.

Have a nice day!

On Sunday 03 October 2004 16:16, Thomas Zander wrote:
> As I explained a year ago, a weight for each filter, and a maximum weight a
> filter-chain can take is the easiest and best solution to this problem.
> With each filter getting an individual weight, its very powerful.
>
> Its not clear to me how your solution would be better (I don't even get
> that it would stop long chains while allowing non-aburt, but very long
> chains)
>
> On Sun, Oct 03, 2004 at 03:30:05PM +0200, Nicolas Goutte wrote:
> > Since the change of the filter system (for KOffice 1.2), the filter
> > manager allows absurd filter chains like:
> > KSpread -> plain text -> KWord -> Latex
> >
> > As replacing again the filter system is going to be hard and as we are
> > moving to OASIS, it is also not much the time for a big rewrite in this
> > code.
> >
> > I had thought long about simple solution and I had the idea of a kind of
> > namespaces (KWord, KSpread...) but I had not any good idea how to
> > implement it until now.
> >
> > Now I have thought about extended mimetypes, like
> > text/plain;kword
> > so that especially text/plain can be differenciated.
> >
> > I think the best would be to add 2 new entries to the .desktop files of
> > filters, adding the namespace at input and at output. (It can be
> > different, see the KPresenter to KWord filter.)
> >
> > As far as I see, this method would stop absurd filter chains without
> > limit any other feature of the current filter systems, like very long
> > filter chains.
> >
> > Also the seems to allow not to need changes to the code of each filter,
> > only the filter system code seems to need changes.
> >
> > Have a nice day!

_______________________________________________
koffice-devel mailing list
koffice-devel@kde.org
https://mail.kde.org/mailman/listinfo/koffice-devel
[prev in list] [next in list] [prev in thread] [next in thread]