"There are still lot's of areas for improvements"

von Redaktion  |  28. September 2005, 08:27

Novell-hacker Michael Meeks talks about the OpenOffice.org development process and necessary changes

WebStandard: Michael, what is your guess: When will OpenOffice.org 2.0 be released?

Michael Meeks: Obviously this is up to the community, so this is difficult to predict. And it has been difficult to predict for quite a while, but for us as Novell it is possible to release earlier and that's what we did, because we think it's good enough. So we took a branch and stabilized that. I think 2.0 has been pretty good for a long time now. That's also the reason why we have shipped pre-releases of it in SUSE Linux 9.3 - and will do again in 10.0, which is quite a stamp of approval.

WebStandard:But the official version should already have been released since quite a while, right?

Michael Meeks: Well, I think it is hard to make a point for 18-24 months release cycles in a free software project, so we are really looking into accelerating that in the future. Six month release cycles really help build momentum and energy.

WebStandard: Do you think the development process of OpenOffice.org is transparent enough or is SUN doing to much behind closed doors in their Hamburg offices? For example, it's really difficult to predict, when releases are going to be.

Michael Meeks: There is some truth in that. But SUN are really trying to do the right thing, in terms of licensing or putting their work back in the public sphere, they have been doing brilliantly. So it is all there, but people don't know about it, there is no way to easily get involved and understand and learn about what's going on quickly. Novell is trying to fix that with the planet and a bug interface that we set up, so that you can get a feel for the pulse more quickly and get involved.

WebStandard: But there is no publicly available roadmap?

Michael Meeks: Yeah, I think that's a problem. But I think while you are stuck in a certain "We are going to release on a schedule that's convenient for us"-mentality, revealing that would also tell people about SUNs internal roadmap. If you go with a six-month schedule instead, it doesn't matter too much, when exactly you release in that timeframe, you still get a stable product. There is also too much "process" in the OpenOffice.org-community, so say if you want to commit something, instead of just committing you have to write a specification, which has to be approved by several SUN people, who are difficult to get in contact with, cause they are busy people. So the process to get a new feature in takes forever, which can lead to it getting tested less. That also explains the rationale behind ooo-build and why we have that sort of proving ground for patches before they go upstream. I think we need to move this whole process elsewhere, so that developers can see, when a release is going to be.

WebStandard: Do you think OpenOffice.org is difficult to get in for new developers?

Michael Meeks: Yes, it's difficult to know where to start fixing. There are some problems, which we are trying to address, like we have our own Bugzilla-interface, but I think soon there will be the OpenOffice.org-Conference in Slovenia and that should be the right time to sit down and discuss this things in person and see how we can fix them.

WebStandard: Recently the license for OpenOffice.org has been changed, what advantages does this offer?

Michael Meeks: Previously there were two licenses, the Sun Industry Standards Source License (SISSL) and the GNU Lesser General Public License (LGPL). The SISSL would allow you to do modifications to the code and under certain circumstances not give them back to the community, and there were a number of companies and people out there who were doing that, which can fragment the community and not help everyone. Given the amount of work SUN is doing it would be right - and better for everyone - to give the modifications back, like we for example do. On the other hand for ISVs it is very important to use the LGPL and not the GPL, so that people can link their applications against it, like one our partners here at Brainshare - Fabasoft - do, and automate in reuse it in their huge piece of software.

WebStandard: OpenOffice.org takes ages to build, quite a time to startup and has a large memory footprint - do you sometimes think it is too big / bloated?

Michael Meeks: I do, but we are getting definitely better in this respect [points to some numbers underlining recent improvements in startup time], it should also be possible to get memory footprint down. The problem is, it takes development time to profile it and this does not show up as immediate progress, but Novell has invested quite heavily in having the right tools for this. Unfortunately at the moment 2.0 is still using more memory than 1.1.x did, but we are going to optimize that.

WebStandard: With the original Ximian-edition of OpenOffice.org 1.x the differences between the vanilla version and your enhanced edition were quite notable, with 2.0 the differences seem to have narrowed down a bit. Would you say this has something to do with upstream now integrating new stuff faster?

Michael Meeks: Yes this has improved a lot now, with 1.1.x and the long release cycle to the next version we had to backport a lot of stuff ourselves as SUN wasn't accepting new features. But with 2.0 I'm optimistic that we are going to have a smaller set of patches. Our focus has really shifted as Novells OpenOffice.org-team has been growing from backporting things to do new work like Cairo-support, the Visual Basic-stuff and Mono and Evolution integration.

WebStandard: Will OpenOffice.org 2.0 build on 64-Bit-systems when it comes out?

Michael Meeks: No, certainly not for 2.0, it's quite a large and difficult problem to get this fixed, a lot of testing has to go into this. There is a lot of interest in this and we're doing some work on it, but I'd say another six months at least.

WebStandard: Lot's of other free software projects like GNOME and Firefox are trying to simplify their user interfaces and preferences options to improve usability. Is this an area where OpenOffice.org still needs improvements?

Michael Meeks: Yes, definitely. There are still lot's of areas for improvements", like the Auto Correct stuff, and there is a mass of options in between it is sometimes difficult to find the relevant ones.

WebStandard: There have been a lot of discussions in the community about OpenOffice.org 2.0 and it using Java more and more, so what's your take on this?

Michael Meeks: Java is great and SUN is using it cause it gives them a cross platform toolkit. That can be unfortunate for free systems as for example if you want to play media you need the java media framework, so others like us are trying to provide more open alternatives. But it's very easy to tell SUN, that they should use something else - they are not burning developer time just for fun, and it simply is the best solution for them.

WebStandard: What enhancements to OpenOffice.org can we look forward after 2.0?

Michael Meeks: I would say the Visual Basic-compatibility-stuff is really good, cause it is really useful for lot's of people. Also further improving performance, is very important, so this is the area which I work on most of the time nowadays. And this is very interesting stuff, cause it really is a wider problem than OpenOffice.org, going through the glibc, the compiler, the whole toolchain to improve that, so that's good fun.

WebStandard: Do you plan to release an enhanced windows version of OpenOffice.org

Michael Meeks: We have a build which we are using internally, but there are no plans to release this publicly. Though the necessary source code to build it yourself is all out there.

WebStandard: Inside Novell quite a bunch of people are employed to hack on "cool" new stuff like Beagle / Hula / F-Spot. Would you sometimes prefer to hack on something which provides more easily visible rewards than OOo?

Michael Meeks: Obviously there are other things one could be doing and these things you mentioned are indeed interesting, but the nice thing about OpenOffice.org is, that it is so big and such a challenge and there so few people that know anything about it and you help so many people by just doing a very small change. So working on it really is rewarding in that sense.

The questions where asked by Andreas Proschofsky during Novells Brainshare in Barcelona
druckenweitersagen:
posten
Posten Sie als Erste(r) Ihre Meinung

Die Kommentare von User und Userinnen geben nicht notwendigerweise die Meinung der Redaktion wieder. Die Redaktion behält sich vor, Kommentare, welche straf- oder zivilrechtliche Normen verletzen, den guten Sitten widersprechen oder sonst dem Ansehen des Mediums zuwiderlaufen (siehe ausführliche Forenregeln), zu entfernen. Der/Die Benutzer/in kann diesfalls keine Ansprüche stellen. Weiters behält sich die derStandard.at GmbH vor, Schadenersatzansprüche geltend zu machen und strafrechtlich relevante Tatbestände zur Anzeige zu bringen.

© derStandard.at GmbH 2010 - Impressum

Alle Rechte vorbehalten. Nutzung ausschließlich für den privaten Eigenbedarf.
Eine Weiterverwendung und Reproduktion über den persönlichen Gebrauch hinaus ist nicht gestattet.