Re: [Conglomerate-devel] Re: Making Conglomerate part of Gnome Office

[Dave Malcolm <david davemalcolm demon co uk>]

On Thu, 2003-09-18 at 02:45, Jody Goldberg wrote:
> On Thu, Sep 18, 2003 at 02:38:21AM +0000, Dave Malcolm wrote:
> > 
> > I think that Conglomerate ought to become a part of Gnome Office.  In
> > fact, Jody gave his blessing to the idea at GUADEC, although this was in
> > the pub, and large quantities of Guinness had been consumed :-)
> 
> Despite being beer-less at this moment I still agree, with the
> standard provision that no app should be released before its time.

Fair enough.

>  
> > Conglomerate isn't yet stable/fast enough for production use, and might
> > not yet be ready for integration into your release schedules.
> 
> Exactly.  Our current target is to sync our release schedules to
> GNOME itself.  The nice solid promises of abi/api stability and the
> time based releases make life much much simpler.  Do we want to
> start defining two classes of Gnome Office elements ?
> 
>     1) stable
>     2) cool stuff that going to be stable

That feels like a good classification to me.  I like to think of
Conglomerate as being cool, but it's not yet ready for production use. 
I've fixed a lot of the most glaring stability problems, but it's now
suffering some major slowdowns in non-trivial use. :-(

> 
> Can you project in rough, non-binding terms your confidence level
> that a 1.0.x would be feasible for GNOME 2.6 ?

Competing against the standard set by AbiWord and Gnumeric: no way! ;-)

Based on the current team (me, plus a few heroic people who help with
packaging, translations, and provide occasional patches):  perhaps.

We don't yet have an agreed-upon feature set that we're aiming at. 
Also, it's _hard_ to offer a 100% stability guarantee for arbitrary XML
editing; there always seems to be another case involving combinations of
entities, external vs internal doc types etc etc. 

I've been too focused on adding cool features for myself, and not worked
enough on getting new hackers involved.  I'm now working on the latter. 
I hope that with a move to GNOME CVS, and some inclusion in the GNOME
Office suite/family/gang that the project will be more visible.  So any
estimates I give will vary according to that.

Perhaps we could label it as a a "Technology Preview" or something when
doing a release to emphasise the fact that it's not yet up to the same
standards as AbiWord/Gnumeric.  It could also pop up a warning dialog
when you load documents that look like they might fail (as well as a
warning document when the app first boots).

We need to have a discussion on the conglomerate-devel list about what
we'd think of as a 1.0.0 release (i've sent an email to that list to
start a thread).

> 
> > At the least, please can Conglomerate appear on this page:
> > http://www.gnome.org/gnome-office/
> 
> There is no polite way to describe how unmaintained and out of date
> that page is.  I'm working on something to post about it shortly.

Ah.  That makes sense...  there seemed to be a discrepancy between the
postings on this list and the stuff on that page.

> 
> > In addition, there are some pieces of code that Conglomerate contains
> > that ought to be useful to other office apps.  The main ones are
> > routines to generate HIG-compliant (I hope) error messages from GnomeVFS
> > errors for File->Open, File->Save, and a Save before Closing
> > confirmation dialog.
> 
> These seem quite nice.  As previously discussed my gut reaction is
> that they belong in libgsf, failing that possibly in libgoffice.
> Information gleaned from their design could feed back into the error
> handling there.

Looks like I'd better have a good look at libgsf.

> 
> > There's a bunch of other stuff as well.  All of this ought to be spun
> > off into some kind of shared library, I suppose.
> 
> We'll need to monitor what goes into libgoffice carefully to help
> ensure that it doesn't turn into a vast dumping ground.  So far its
> maintained its simplicity.  As a urologist friend used to say
>     'drop your drawers and put your problem where I can see it'
> Lets go through on a case by case basis and hash out what makes
> sense.  The more code we can share, the more polished our apps can
> be.

Does libgoffice exist yet?  I couldn't see it at:
http://cvs.gnome.org/bonsai/rview.cgi?cvsroot=/cvs/gnome

> Do we need a quorum of some sort before smearing conglomerate with
> sacred gnome droppings ?

Now you're scaring me. :-)



Dave