Re: GNOME Office and OpenOffice

[Martin Sevior <msevior mccubbin ph unimelb edu au>]

Hi everone,
	   Here are my comments to Havoc ALMOST uncensored :-)

Martin

>From msevior mccubbin ph unimelb edu au Thu Oct 26 08:53:55 2000
Date: Wed, 25 Oct 2000 21:35:34 +1000 (EST)
From: Martin Sevior <msevior mccubbin ph unimelb edu au>
To: Havoc Pennington <hp redhat com>
Subject: Re: Comment requested on GNOME Foundation



On 24 Oct 2000, Havoc Pennington wrote:

> Let me reiterate that if Abi or parts of Abi are the best technical
> choice, they definitely should be used, and it's my expectation that
> they will be used.
> 

OK Havoc this is just between you and me. How on earth are you going to
combine bits of Abi or Gnumeric with bits of Open Office? Gnumeric is
written in C, Open Office is C++. Abi is GPL and can't be added to Open
Office since SUN wants code in Open Office to be dual licensed. Even if
the Hackers in Gnumeric agree to a license change, you can be sure that
Abi hackers won't. Abi and Gnumeric can raid Open Office for code but
not vice versa. 

So the only practical thing you can do is add code to Abi and Gnumeric
unless someone starts from scratch. Is that the plan? If so no one has
told us.

I have seen no indication that anyone from SUN (apart from Coalan) is the
slightest bit interested in Abi anyway. I wouldn't know about Gnumeric
which in my opinion is already a better program than the Open Office
SpreadSheet.

> There is a huge, huge task ahead of us; we have quite a large GNOME
> Office codebase, and an even larger OpenOffice codebase, neither one
> is finished, we have a couple hundred people wanting to work on this
> stuff living all over the world, somehow it has to all get organized
> to deliver one full-featured office suite. Not going to be easy. 

Well you need a framework to code on. In the case of a WP you have a
choice of Abi or Star. Maybe this is my lack of imagination? Do you see a
different way to do this?

Here is my set of comparisons between Abi and Star.

1. We use 16 bit words to hold our characters. We can easily do full
internationalization. We have just integrated Russian support into CVS
and will integrate Chinese Real Soon Now. There is already a proof of
concept patch to do this. CKJ support will quickly follow.

2. We can do Modeless dialogs. After 10 years SO can't.

3. We already use Gtk for our GUI. 

4. We already use lots of Gnome Libs and more will be added Real Soon Now.

5. We are likely to get Bonobo going shortly (its next after gnome-print).

6. Our memory footprint is a tiny fraction of Star. We have a shot at
being used in embedded devices. Abi is the second most downloaded app for
QNX.

7. We have an established developer network already.

8. Our License is GPL. How many Gnome hackers will want to dual license
their code? 

9. Our code is lot easier to understand for your average hacker.

On the other hand OO has 50 full time hackers and maybe has paid other
gnome developers to work on OO. They have to work from a 6 million line
legacy code base a lot of which will have to re-written just to get the
features we have now. It also has the backing of a 200 Billion
dollar company. Geeze that's 2/3 Australia's GDP!

Given all this it appears to me that the best way to get a GPL'd state of
art WP for Gnome is to add features to Abi. The only reason there is even
a question about this is the money and hackers SUN is offering. Even so
given the dual license bit how can Gnome in all honesty recommend to your
average hacker that they contribute to Open Office?

 > Since
> the OpenOffice code just went public a week ago, it's a bit early to
> say how it's going. But I'm optimistic that it can be done; so far Sun
> seems to be earnestly trying to get their developers participating in
> the GNOME community, and we already have the first wave of patches
> from them.
> 

I can see we have a collaboration on infrastructure things like bonobo and
gnome libs. I can't imagine Gnome agreeing to devote resources to develop
code under anything other than GPL or lGPL.

I'm sorry I guess I lack the imagination for how we are practically going
to combine numerous incompatible pieces of code to make a coherent whole.

Sorry for my focus on Abi but it is the only piece of code I'm familiar
with and I have thought about this a lot.

Best Regards,

Martin