RE: [java-gnome-hackers] Looking forward

[Luca De R. <pie...@fa...>]

Il mar, 2004-04-06 alle 13:51, Jeffrey Morgan ha scritto:
> > > 1) 
> > I think keeping in sync with native libraries and fixing bugs 
> > should be
> > our top priority - it should be ok as long as we stay on top of it. 
> > 
> > In terms of fixing bugs, how long do you think we should wait before
> > creating a new cvs branch to separate stable/unstable releases? I
> > suspect we will have a lot of new people using java-gnome and finding
> > new bugs. Backporting fixes to each branch is hard work, so 
> > I'd suggest
> > we leave it as long as possible before making a branch. 
I agree with Mark: do you remember how many bug reports we have had when
we released 0.8.2 (and joined the official bindings)?
We couldn't take care of our users whining for a 0.8.3 release, cause we
were in business with 2.5.3.

> I personally would like to make the branch soon after 2.6 because I
> wish to attach the 64-bit issue and really shouldn't do this in a
> "stable" branch.  A question we should ask ourselves is "should we only
> provide bug fixes for the 2.6 release and forward or should we also
> provide bug fixes for the 0.8 branch?".  We also need to agree on a process
> we can use to keep the stable branches up to date.  Please let me know
> what you think.
I think, maybe, we should look more to the practical aspect...(really no offense intended:))
We should ask ourselves: do we need 64-bit support? Who is going to use
it in the near future? Is someone managing to build a useful app with
it? (well not that I want a promise of faith but maybe someone could
have been hired to write something with it (like Bob Fischer with the
iPaq...))
What I say is that if we are going to add 64 bit support only because
having it is fine, perhaps we could work on the existing code to fill
the holes that are present in the bindings, instead. (like DnD...)
In this way, I hope we'll see more applications spreading out, cause our
users will have a stronger and more complete library to play with.

My point is that it's better to keep fixing bugs for the 2.6 release
while our users will ask for them. Then we could accept patches for the
0.8 branch but we should explicitly encourage people to use 2.6, because
of the large changes that have been done.

I'm tempted to say that Jeff, Mark and Jonas could take the devel
branch, and work on adding new modules , on keeping in sync with
gnome,and on adding 64bit support (if it's really needed), because this
is the more work, while the others and I should maintain the stable
branch (with a lot of help from you of course...:)).
But since I understand that, at least Mark and I are a bit under
pressure with the uni, I don't know if this idea could fit at the
moment...
So maybe the best solution now is to wait until the gnome-2.7 release
cycle to add new modules and 64bit support. In this way we can start
from a heavily fixed version of j-g-2.6.X without the need of
backporting anything. When the 2.7 branch will start, then I could take
care of backporting eventual fixes in the 2.6, but since it has just
passed a six months bug-busting period, hopefully there will be few bugs
to fix between 2.7 and 2.8.

> > > 2) Lazy Event Registration - Both Jonas and Mark have suggested this
> > > change.  It makes a lot of sense and should be a nice performance
> > > improvement.
> > IMHO, the only performance problem java-gnome has is the start up time
> > of the jvm, which we can't change. So I'd say this might be lower
> > priority. 
> 
> I think this is more than just a performance fix.  It is a much cleaner
> implementation and should be a simple change to make.  Since this
> change impacts many classes in the project I personally would like to get it
> out of the way early in the development cycle.  Perhaps Jonas can implement
> this item.  What do you think Jonas?
> 
> > libgnome-panelapplet has been requested a few times. 
> 
> This will require Bonobo support.  Adding Bonobo to java-gnome will be a
> fairly large undertaking and I was hoping to put it off until the next
> release cycle.  If everybody feels this is more important then we can
> postpone gnome-vfs support until later and start work on Bonobo and then
> the panelapplet.
> 
> > I would also like to see full drag and drop support for treeviews. It
> > might be possible now, but AFAIK, it hasn't been done.
> 
> I think we need to appoint somebody in charge of drag and drop.  It
> does not work throughout the bindings but I believe it is very close
> to working 8-)  Somebody please take this task.
I would take care of fixing the stable branch, and I think this implies
also implementing DnD...but are you talking about to get it fixed
*before* we release final? Because, unfortunately, I will find a bit of
time only after 17th.

I hope my thoughts weren't much confusing, and sorry if something come
across wrong: I only want to help, really!

OT:did anybody watch "revolution os"? I think it's a "must" :)
-- 
Luca De Rugeriis <pie...@fa...>