incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From robert_w...@us.ibm.com
Subject Re: OpenOffice Apache Incubator Proposal and uniting "The Community"
Date Fri, 03 Jun 2011 00:22:59 GMT
"Allen Pulsifer" <pulsifer@openoffice.org> wrote on 06/02/2011 06:58:45 
PM:

> 
> As a long time member of the OpenOffice.org community, I would like to 
offer
> my thoughts on the Oracle/IBM proposal.
> 

Thanks.  This is a great summary of the history.

.
.
.


> 
> Despite the fact the IBM's vision for OpenOffice seems to correspond 
pretty
> closely to the Apache vision or philosophy, it should be noted that 
there
> are some apparent differences between OpenOffice, in its current or
> historical form, and other Apache projects.  First, while I am most 
familiar
> with the Apache HTTP Server, I get the sense that most Apache projects 
are
> intended for system administrators or software developers, both of which 
are
> technical folk.  OpenOffice does not currently fit that model.  In its
> current form, OpenOffice is a desktop application, not a server or a
> software development toolkit.  It is used by primarily non-technical
> end-users in an office or personal setting, not by technical people.

Well, yes and no.  Some actually do use OpenOffice as a server 
application.  For example, Apache is currently tracking another proposal 
for a project that uses OpenOffice as a server:

http://wiki.apache.org/incubator/OpenmeetingsProposal

But it is important to acknowledge that this is primarily a mass-market 
end-user application, and we'll probably want to have an end-user portal 
for downloads, documentation, templates, tutorials, etc., as well as a 
project portal. 

My understanding is that Oracle will be contributing the domain name 
openoffice.org to the project.  I think that would be a good basis for the 
end-user site, since that is the destination of 220,000+ links.  The 
standard Apache infrastructure should be fine for the project site.


> Second, if Apache takes on OpenOffice, it will instantly become the 
Apache
> project with the largest number of end-users (if you include the number 
of
> users of both the Apache version and its derivatives).  In fact, the 
number
> of users of OpenOffice may be larger than all other Apache projects

Apache already hosts 5 of the most downloaded open source projects.  From 
what you are saying, they may soon have 6 of 10.

> combined.  Third, OpenOffice historically had (I think) a much larger
> community around it than any current Apache project.  This community has
> historically included people doing coding, localization, Q/A, 
documentation,
> marketing, distribution and end-user support.  While other Apache 
projects
> may have had some or all of these activities, I think it may be correct 
to
> say that the size of the community around OpenOffice was historically 
larger
> than all Apache projects combined.  Another difference I can see is that 
the
> OpenOffice codebase is huge, complex and "old".  It has been under
> development since 1984, approximately 10 years longer than the Apache 
HTTP
> server.  Its source tarball is approximately 250 MB.  It has its own way 
of
> doing just about everything, from its underlying object model (called 
UNO)
> to its GUI framework.  Very few developers understand the code.  The 
code is
> also poorly documented, with much of the documentation disorganized,
> incomplete, unclear, out-of-date, and/or simply wrong.  Making changes 
to
> the code is often a frustrating exercise in decoding several layers of
> unnecessary abstraction spread across multiple source files, and then
> digging in with multiple trial-and-error changes while generally 
breaking
> more than you fix.  All that is to say, OpenOffice is a very big baby to
> adopt, and will have unique and significant needs.  It may result in a 
huge

A slightly different perspective.  We formed a team in Beijing, with no 
formal knowledge of the code. Most of them were fresh out of college. They 
were a very young team at the time.   With a little help they got up and 
running.  They did some ambitious changes, integrated OpenOffice into 
Eclipse.  The first version was very rough.  But they kept plugging at it. 
 Yes, having code comments in German was not particularly useful for a 
Chinese C++ programmer.  But they didn't let it get in the way.  I think 
you can use the LibreOffice example as another "existence proof" that it 
is possible for new developers to get up to speed in this codebase. 


> influx of people looking for all sorts of different things.  It is not 
clear
> the Apache's existing infrastructure is sufficient to handle a project 
or
> community of this size and complexity, and it may have to be upgrade 
with
> new tools.  It would not surprise me if adopting OpenOffice did not 
result
> in eventual changes in the Apache's organization and direction, possibly 
in
> unpredictable ways--not that that is a bad thing, it's just something to
> think about.
>

The only concrete thing I've heard so far is the question of whether 
subversion can handle the project.  Are there any other specific aspects 
that you think we should watch out for?

For mirroring, I see 295 mirrors:

http://www.apache.org/mirrors/

Current Apache distribution is said to be 32GB.  So the 250MB tarball is < 
1% increase.

I suspect the end-user facing web site will need capacity.  I'm seeing 
estimates of 1.3 million monthly unique visitors.

.
.
.

> My question is simple: has IBM consulted with the companies that are
> currently paying developers to work on LibreOffice to determine what 
they
> want to do?  Can these companies foresee paying developers to work on an
> Apache Licensed version, or to make contributions to an Apache Licensed
> version?  If so, do they also foresee paying developers to continue to 
work
> in parallel on an LGPL version, and if so, how do they foresee these two
> projects being coordinated?
> 

We're in the process of talking to a number of companies.  It would not be 
proper for me to name them. 

As for how the projects can coordinate, clearly we know of one model that 
could work.  That is the same model used previously when Sun's 
OpenOffice.org project was the central repository and satellite projects 
contributed.  That model is morphed in Apache in several ways, which I 
think are positive changes:

1) No copyright assignment required by contributors, so rights are all 
symmetrical

2) No corporate gatekeeper determining which upstream contributions will 
be accepted. We'll hash out any disagreements like this, in public, on the 
dev list.

3) Apache 2.0 license will be more attractive to many corporate sponsors 
of possible project developers. 

Regards,

-Rob

---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@incubator.apache.org
For additional commands, e-mail: general-help@incubator.apache.org


Mime
View raw message