[tdf-discuss] IAccessible2, IBM, Oracle (was: (...) Proposal to join Apache OpenOffice)

Hi Christian, All,


This is my fifth attempt to send a response to 
the list (after previous attempts on 5, 8, 15 and 
26 July). I hope it gets through this time.


At 18:21 5-7-2011, Christian Lohmaier wrote:
> Hi Christophe, *,
>
> On Tue, Jul 5, 2011 at 4:50 PM, Christophe Strobbe
> <christophe.strobbe@esat.kuleuven.be> wrote:
> > At 16:14 5-7-2011, Christian Lohmaier wrote:
> >> On Tue, Jul 5, 2011 at 3:34 PM, Christophe Strobbe
> >> <christophe.strobbe@esat.kuleuven.be> wrote:
> >> > At 23:16 4-6-2011, Christian Lohmaier wrote:
> > [...]
> >> Well, as seen on this list (by Malte's post), apparently there has
> >> been work on a *private* cws that nobody in the community (and yes,
> >> people who are working on private cws are not part of "the community"
> >> in this regard - they are of course for that part of their work that
> >> happens in public)
> >> All promises IBM is making/has made so far is only lip service for me.
> >> I only believe it after I see the actual contributions from them.
> >> (And as written I don't consider code dumps that need a man-year of
> >> work to get integrated as contribution)
> >
> > If Oracle asks IBM to implement IAccessible2 on version 3.1 and releases
> > OpenOffice.org 3.2 before IBM has submitted 
> the IAccessible2 implementation,
> > how is IBM to blame?
>
> Reality check please. 1st of all: What is stuff you know, and what is
> stuff you guess?
> Do you know that the 3.1 based ia2 dump/work is because Oracle asked for it?
> If Oracle asked for it, do you know when Oracle asked for it?
> Do you think Oracle really is so stupid to explicitly ask for code
> based on an old branch?
> If Oracle did ask for it, and IBM did "contribute" - why wasn't the
> cws integrated?

CS:
What reality check? I talked to and exchanged mails with people in
Oracle and IBM. Since September 2008 I have been involved in a European
R&D project on accessibility where Oracle (previously Sun) is one of the
biggest partners. So I talked to accessibility folks at Oracle. That
is reality.
Yes, IBM donated an IAccessible2 implementation to Oracle. Malte
Timmermann talked about this at FOSDEM 2011:
<http://www.fosdem.org/2011/schedule/event/ooo_accessibility>.

Just like you I don't like secret deals about open source projects.
I agree it is better if code contributions should happen on public
repositories. But that does not mean these things don't happen.


At 18:21 5-7-2011, Christian Lohmaier wrote:
> 2nd) Obviously you cannot integrate something that is not ready.
> Why was it not ready? Because nobody worked on it.

CS: I throw your own question back to you: "What is stuff you know, and
what is stuff you guess?" ;-)


At 18:21 5-7-2011, Christian Lohmaier wrote:
> Who could do the
> work on it? Of course best the developers who know the code, i.e IBM
> developers.
> And you cannot delay a release for years. (the cws Caolan mentioned in
> the blog-comment was created in 2010-05 - while the branch-off for 3.2
> already happened 2009-09 more than half a year earlier)
>
> > Between 3.1 and 3.2 the code had changed and had been moved to another
> > type of repository.
>
> Again reality check. Oracle surely did ask for the code to be
> contributed against the current, actively being-worked-on codeline.

CS: Again: "What is stuff you know, and what is stuff you guess?"
Oracle had released a newer version by the time IBM submitted their
code. Fact: this was stated in public at FOSDEM 2011.


At 18:21 5-7-2011, Christian Lohmaier wrote:
> A
> codeline that is not in feature-freeze. What IBM then delivers is a
> completely different question. Also whether Oracle/Sun asks for it in
> 2008, but IBM delivers in 2010, it's obvious that code makes progress.

CS: Obviously. Nobody is debating that code makes progress.
I don't know when Oracle asked for it.
IBM promised to add IAccessible2 in 2007
"IBM joins OpenOffice.org community - will contribute IAccessible2 support":
<http://blogs.oracle.com/korn/entry/ibm_joins_openoffice_org_community>
(10 September 2007).
A few weeks later IBM wrote they were adding it: "IAccessible2 is in or being
implemented in these products today: (...)
* Open Office (IBM is just beginning the contribution effort)":
<https://www.ibm.com/developerworks/mydeveloperworks/blogs/schwer/tags/windows?lang=en>
(27 September 2007).
In July 2008, people said it that it shouldn't be expected before the end
of 2008 (OpenOffice.org 3.x but not 3.0):
<http://www.freelists.org/post/nvda/nvda-and-openoffice,13>.
The contribution was mentioned again at teh OpenOffice.org conference
in November 2008 (in a talk by Michael Karasick, Director of Lotus
Development IBM China):
<http://blogs.oracle.com/malte/entry/iaccessible2_or_ibm_to_contribute>.



At 18:21 5-7-2011, Christian Lohmaier wrote:
> > That is the reason for the complex and time-consuming
> > integration work that Oracle needed to do for IAccessible2.
>
> NO! Why does it have to be Oracle to do the integration work. Again
> one of the points about collaboration. Just uploading a
> million-line-codepatch somewhere is not contributing. It is complying
> with whatever deals that were signed or to comply with license matters
> at best.


CS: Yes, the collaboration you write about is what *should* happen,
but that is not the version I heard from Oracle.


At 18:21 5-7-2011, Christian Lohmaier wrote:
> > The integration and testing were still in progress when Oracle decided
> > to stop investing in OpenOffice.org. As far as I know, that is why
> > the IAccessible2 code did not end up in public repositories.
>
> Again this is stupid argumentation. We're talking about a OpenSource
> software here after all.

CS: You can call the argumentation stupid, but the integration and testing
*were* still in progress when Oracle decided to give up on OpenOffice.org.
Don't blame the messenger.

[Aside:
OOo was not exclusively open source, if version 1.1.4 was dual-licensed under
LGPL and SISSL to enable IBM to base Lotus Symphony on it without releasing
changes to the source code. Sun withdrew SISSL but later relicensed OOo to
IBM under a private agreement (see top of
<http://www.h-online.com/open/features/OpenOffice-splits-and-pirouettes-1270296.html?page=2>).]


At 18:21 5-7-2011, Christian Lohmaier wrote:
> And we're not talking about weeks, but years.
> We're talking about big announcements to dedicate more than 30
> developers to work on the officesuite and collaborate with upstream,
> but no results after 4/5 years.
> And this further proves my point about questioning IBM's commitment.
> Lip service, but no actual work that ends up "upstream".
> They did not contribute to OOo, but they did drop some code at Oracle.
> Again this is not my idea of contributing to the project.


CS: I agree that this is not how open source development should be done.
My intention was not to defend this but to highlight the importance
of IAccessible2.


At 18:21 5-7-2011, Christian Lohmaier wrote:
> > The contribution to the 1.1.5 codeline is irrelevant because completely
> > outdated. I added that note merely as backgound information.
>
> No, it is not irrelevant, because it is the very same situation. Big
> announcement "we will conribute, we have lots of manpower" but no
> results. That's the whole point. IBM doesn't have a record of being a
> good contributor, the opposite is the case. And to change this, we
> don't need another lip-service announcement, but actual code
> contribution.
> That you can only point at Ia2, but not at other work is further prove
> of this topic.

CS: That does not prove anything. I have no overview of the code or its
contributors.

Just like you, I don't like big announcements that lead to few concrete
results. The intention of my original response was to draw attention to
IAccessible2, which is a valuable contribution in its own right, even
though you clearly expected more from IBM.


At 18:21 5-7-2011, Christian Lohmaier wrote:
> And don't get me wrong, I'm sure that you'll see IBM contributing to
> apache-OOo, at least until you can actually build something from
> Apache-OOo sources you can ship to the users, but after that I'm
> pretty sure that IBM will focus again on its very own Symphony and
> only do the necessary stuff to keep their own stuff compatible.
>
> And don't get me wrong�: I'd be happy if IBM proves me wrong.
>
> > If Sun/Oracle engineers state that IBM donated the IAccessible2
> > implementation, it is unlikely that this piece of work was done
> > by Sun/Oracle.
>
> Again it is not about the Ia2 work itself, but the porting from the
> old 1.1.5 codedrop to "current" codeline.
> You apparently don't know any hard facts about this, neither do I. So
> while you claim that Oracle did ask IBM for the code ported to the 3.1
> codeline, and that IBM then followed this request, I question this
> scenario.


CS: Instead of questioning my credibility, you could ask people at
Oracle and IBM who were involved in this. That's what I did. And you?


At 18:21 5-7-2011, Christian Lohmaier wrote:
> Or even if IBM did contribute it against the 3.1 codeline: Why is it
> still not integrated? This can only mean that a huge amount of work
> is/was still left.
>
> You mentioned that the vcs moved between 3.1 and 3.2 - but sorry
> again, this is /no/ technical hindrance to switch from one vcs system
> to another. The code is the same, no matter what. And you got the very
> same "branch points" to base your work on in the new VCS system.


CS: IBM worked on version X that came from vcs system A; when they
donated the code, OOo was at version Y in vcs system B. Consequence:
it took time to find out what code should go where.
That's what I heard at FOSDEM in February. Interpret it as you like.


At 18:21 5-7-2011, Christian Lohmaier wrote:
> >> > At this moment I know no one at Oracle who can or wants to say how much
> >> > of
> >> > the IAccessible2 implementation will end up in OpenOffice.org 3.4.
> >>
> >> Well, then you missed Malte Timmermann's post.
> >
> > Yes, I missed that. (Curiously, he sent that message from a private
> > address, not an Oracle address.)
> (...)


At 18:21 5-7-2011, Christian Lohmaier wrote:
> >> (about the status of
> >> iaccessible2), As Rob is strongly against releasing OOo 3.4 with the
> >> "blessing" of the apache-OOo project ("take that discussion to the old
> >> OOo-lists" basically (paraphrased)), I doubt there will be a OOo 3.4.0
> >> at all.
> >
> > If that is true, that will be a loss for the 
> accessibility of OpenOffice.org
> > and LibreOffice on Windows.
>
> The project did wait 5 years for it, it can wait another two...


CS: If you think it's OK for blind users on 
Windows to wait another two years [1],
I'll start recommending IBM Lotus Symphony to them. It is much more accessible
than LibreOffice or OpenOffice.org (since you question my credibility, don't
take my word for it, ask blind users!).

Best regards,

Christophe Strobbe


[1] "We do not want to wait, why should we, its been too long already.":
   <http://www.freelists.org/post/nvda/open-office-accessibility,28>