On 19.04.2006 10:02:19 Chris Bowditch wrote:
> Jeremias Maerki wrote:
> Hi Jeremias,
> > Hi Patrick,
> > 
> > yes, I've seen that something's going on in that corner. After my
> > experience last year, I was a bit hesitant to take another stab, but
> > since we've already worked together, I'd be fine mentoring a little
> > project. There are a lot of things that could be approached:
> > http://wiki.apache.org/xmlgraphics-fop/FOPProjectTasks
> That Wiki is quite old now and a lot has changed since it was written.
> > 
> > The thing that would make the most people happy would be implementing
> > auto table layout. That's something that is not on any of my clients'
> > wishlists so it goes on my private task list on which I have other
> > higher priority items. :-o You'd be the hero implementing that!
> table-layout="auto" is a often requested feature and as Jeremias said 
> I'm sure a lot of folks would be grateful for it. My only concern is the 
> amount of work involved, it might be too big a chunk of work for 
> Google's summer of code project.

Not if we're only talking about the basics. See my other reply. I
estimate that I could probably do the basics in about 4 or 5 days, but
obviously I already know my way around. The most complex task here would
be to split up the generation of the inline element lists from the
line-breaking process which is currently one combined task in
LineLayoutManager.getKnuthElements(). This is because you need the
inline element list to determine the column widths. After you defined
the columns widths you can do the line breaking. No witchcraft here.

> Another broken table feature that often causes confusion is the 
> border-collapse algorithm. I know Jeremias had some ideas about how to 
> implement most of it and just leave out the hardest bits.

Yes, that would also be an option. We simply need to be clear that with
the layout-related tasks work-in time will be longer since you have to
build up some knowledge about the Knuth element model and the XSL-FO
specification (applies to auto table layout, too, but not to the image
package redesign). The table border model is particularly complex
because of the possible interactions of which, as Chris says, some can
be ignored for simplicity for now.

> <snip/>
> Chris

Jeremias Maerki

Reply via email to