[sugar] A few more toolbar buttons for AbiWord?
Mon Nov 20 19:39:31 EST 2006
On Mon, 2006-11-20 at 17:43 +0100, J.M. Maurer wrote:
> On Mon, 2006-11-20 at 11:31 -0500, Ivan Krsti? wrote:
> > J.M. Maurer wrote:
> > > That wouldn't be a requirement with abi collab. We can aggregate lots of
> > > strokes (or better, in abi terms: ChangeRecords, thing "patches to the
> > > document") if needed.
> > How does this deal with conflicts?
> Most of the conflicts can be resolved by recalculating the document
> positions that the characters should go in. The algorithm is pretty
> similar to the one used in darcs.
> > > We have that already sorted out for n people. Even with proper
> > > undo/redo.
> > Is there a design doc somewhere for the way your collaboration works, or
> > do I need to go read the source?
> Martin did the last revision of the code. We did have a design document,
> but I think since the last revision it is out of date.
> About undo/redo, let's quote Martin from one of my blog entries:
> "It handles the case people making changes to other peoples text. We do
> this by detecting that it has happened and preventing undo through the
> remote user's changes. You can't undo a remote user's changes. Just your
> own. If there is a conflict undo/redo prevents further undo/redo's.
> However you are perfectly entitled to manually type over the remote
> users changes or delete them etc."
> > I'm very skeptical of things being as
> > simple as you make them sound, because a colleague recently researched
> > most of the published research on collaborative editing techniques, and
> > found many of the problems I list to be generally still unsolved.
> It's not like we just implemented this within a week or so. We've been
> working on this since June _2005_.
> Now if Martin could drop in to discuss the latest changes :)
I'm back from traversing the distance from Japan to Australia.
Getting full and proper collaborative editing has certainly been
I invite you all to try out our build once Marc finishes the changes to
Regarding the various levels of skepticism about our claims, fair
enough. We have working code however it has not been extensively tested
against a large collection of people working together.
I think the algorithms will scale as they are based on guaranteeing
pair-wise synchronization between users but I haven't tried to "prove"
this. We await large scale testing to find out.
I'll try to update the documentation regarding the algorithms but for me
there is a definite trade off between writing and producing new code and
often prefer just to write new code.
More information about the Sugar-devel