[prev in list] [next in list] [prev in thread] [next in thread] 

List:       koffice
Subject:    Library for MSWord filters
From:       "T.J. Mather" <tjmather () maxmind ! com>
Date:       2002-07-11 2:25:26
[Download RAW message or body]

This posting is to follow up on a couple of threads on how we should 
proceed for writing Word and other MSOffice filters.  We've identified 
two libraries that look promising for this purpose, wv2 and POI.  The 
question is do we want to finish development of wv2, or combine efforts 
with the POI project with ideas from wv2.

Here are the major issues with wv2 and POI:
1. License
	QT is licensed under the QPL/GPL.
	KDE is licensed under the LGPL.
	wv2 is licensed under the LGPL
	POI is licensed under the Apache license.

	Clearly it is OK to use wv2, since GPL+LGPL+LGPL is allowed.

	As for POI, it is not as clear, but it may be OK, since QT is 
	dual licensed under the QPL as well as the GPL.  I believe that
	QPL+LGPL+Apache is allowed.

	Edge: wv2 is winner here for using the LGPL license instead
	of the Apache license.  This is because Abiword is licensed
	under the GPL, so they can link to wv2 instead of having to
	write a standalone filter with POI[1].

2. Dependencies
	wv2 has some minor requirements like glib/mini-glib, iconv.
	POI has a more difficult requirement, GCJ 3.1

	Edge: wv2 is clear winner here.

3. Cooperation vs Competition
	POI has several committed developers, one who is actively working
	on HDF.  Another POI developer is planning to work on HDF too.
	Jakarta hosts some of the best and most widely used Java projects
	and it is reasonable to expect that the HDF project will attract
	a lot of users.

	Working on a MS Word access library is tedious and difficult
	work.  By pooling resources we can make a much better library
	with less effort than we can by competing.

	Edge: POI is the clear winner here.

So, assuming that 

1. QPL+LGPL+Apache is OK
2. The design of the POI HDF library is sound and will allow us
   to add support for WMF and MSOD.

then, as a great over-simplification, basically this comes down to a 
tradeoff between:
Dependency on GCC 3.1 vs. pooling efforts.

-TJ

____________________________________
koffice mailing list
koffice@mail.kde.org
To unsubscribe please visit:
http://mail.kde.org/mailman/listinfo/koffice
[prev in list] [next in list] [prev in thread] [next in thread]