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

List:       koffice-devel
Subject:    Re: save a kword file as a spread sheet
From:       zander () kde ! org
Date:       2004-10-27 8:54:56
Message-ID: 20041027085456.GB5118 () factotummedia ! nl
[Download RAW message or body]

[Attachment #2 (multipart/signed)]


On Wed, Oct 27, 2004 at 10:41:56AM +0200, Werner Trobin wrote:
> On Wednesday 27 October 2004 10:20, Nicolas Goutte wrote:
> > On Wednesday 27 October 2004 09:11, Tomas Mecir wrote:
> > > On Tue, 26 Oct 2004 18:53:38 +0200, Nicolas Goutte <nicolasg@snafu.de>
> >
> > wrote:
> > > > > This is the known "going via text/plain in the filter chain" problem.
> > > > > Nicolas Goutte was looking into it afaik.
> > > >
> > > > Yes, but as fixing it exceed the time that I can guarantee to work on
> > > > it, I have done nothing for now. However I can try to explain to
> > > > anybody wanting to try to fix it.
> > >
> > > Hm, wouldn't it be sufficient to simply add a flag for each
> > > import/export filter, that would state whether output of this
> > > particular filter can be used as a source for another filter? Then set
> > > that flag for filters that output to HTML and text/plain... Maybe not
> > > perfect, but it should work.
> >
> > No, we have decide to do it by filter weights. However that means that new
> > code must be added. (The filter weights are only user when creating the
> > filter chain to process the document and not before showing the file
> > dialog. So we need to transform the code for the file dialog to use filter
> > weights too. I suppose that both codes could perhaps be merged as one but
> > that is to be seen in details.)
> 
> Merging the code is not really possible, IMHO. The two problems we have to 
> solve are:
> - Which mimetypes of the graph have a path to the target mimetype? (for the 
> filter dialog box)

But I proposed to do exactly this using the filter-weights as Nicolas points
out.  Theoretically speaking; solving the this problem will still lead to
having paths between illogical nodes if you don't limit the amount of hops
it can take.

> The two algorithms I implemented are a BFS on the graph with edges reversed 
> (we are searching from the target mimetype towards the sources, i.e. in the 
> "wrong" direction) and Dijkstra's shortest path algorithm (to determine the 
> shortest path from S to T).

The merging of code that Nicolas proposed seems to be about some different
parts then this though; but I don't know the code that well to talk about it.

-- 
Thomas Zander

[Attachment #5 (application/pgp-signature)]

_______________________________________________
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]