Interview: Ed Peterlin and Dan Williams:
OpenOffice.org for Mac OS X
2002-07-25
Introduction
The first Mac OS X port of OpenOffice.org (638c) was released at Apple's World Wide Developer Conference, on May 7. OpenOffice.org was there in strength: we had a very popular booth and Ed Peterlin presented an exciting demonstration of the build before an audience of Open Source developers. (More on this soon.)
In the months since then, the two community contributors primarily responsible for the Mac OS X port, Ed Peterlin and Dan Williams, have substantially furthered their work; although it is still a developer release, it now runs using OpenOffice.org 1.0 and can be immediately downloaded. It also, with some limitations, saves files and prints. The following is a very slightly edited e-mail interview conducted over a few days last May. The build Ed and Dan refer to in conversation is the older build, 638c. If you are curious about what the updated port (1.0) does and does not do, Dan prepared an excellent FAQ.
The People
I have never met Dan Williams; but I did spend a lot of time during WWDC with Ed Peterlin, and I would love to have the talent to put to paper his energy, enthusiasm, and dedication to this project. Both Ed and Dan don't seem to sleep; OpenOffice.org for Mac OS X is their hobby.
Ed Peterlin
Ed's day job is at BIOPAC Systems, Inc., where he is responsible for Macintosh development. In his words: "I write software for displaying and analyzing physiological signals, such as heartbeat and muscle response, that are acquired from our MP hardware." But what drew him to OpenOffice.org?
"I grew interested in OpenOffice.org through using StarOffice on Solaris boxes a few years ago. Having no other office solution for a Solaris machine, I became quite impressed with its capabilities and usability. Even at BIOPAC I've used OpenOffice.org when we ran out of licenses for MSFT Office but had to read technical documents that were written in Word. When I heard an OS X port was released to the open source community, I felt interested in bringing this solution that had proven so useful to me in the past to one of my favorite platforms. I joined the OpenOffice.org project the weekend after I got a machine capable of running OS X, around July-September of last year.
"I became interested in Macs as an extension of the old Apple IIe, my first computer. I was in an Apple-based school system, and after I outgrew my IIe a Mac was the next best step. Not to mention that at the time I could get a IIe emulator card and still play my favorite video games."
Dan Williams
"By day I am an undergraduate Anthropology major specializing in Archaeology (screw Cultural [Anthropology], Archaeology is where the real fun is), which is partially where I get the time to work on OpenOffice.org. I have fun making databases that store and analyze spatial data, and have an active interest in Geographic Information Systems and developing their potential in Archaeology.
"After hearing a lot about StarOffice, trying it out on Linux, and seeing it as really the only viable alternative to Microsoft Office on any platform, I became interested in OpenOffice.org. I looked at it, said, This is cool. It had everything I thought should be in an office suite, and I decided that MS Office wasn't anything special after seeing how good StarOffice was. I thought that, since it wasn't on Mac and I had Unix and Mac OS programming experience, I'd help it get there.
"I got my start with an Apple IIe back in early elementary school, graduated to a Mac LCIII, bought Think C back when it wasn't a Symantec product, and taught myself Mac programming though books. It was amazing that I could create programs that did stuff like professional programs did. From there I kept buying copies of CodeWarrior and keeping up on new Toolbox stuff. When I first used Unix I was impressed, but the Mac was still better to use. When Mac OS X came along, I got very excited because of the power of its Unix base. And because StarOffice was on Unix, it could be on Mac OS X too."
About the Port
Ed: "The build [638c for Mac OS X] was the first of many milestones that the Mac OS X porting effort reached. For the last year, everyone working on the Mac OS X porting project has been staring at error messages coming out of the development tools, and the 638c developer build represented the first time we'd been able to get OpenOffice.org to launch.
"I have been impressed with just how much of the software appears to be functional, a real testament to the engineers who originally designed OpenOffice.org with the foresight that it would be moved into new platforms. Even while watching people play with the demo machine at WWDC I was continually amazed to see just how much works. That said, there's still quite a ways to go.
"Our Quartz native graphics layer is not yet complete, so we can't run the first build on Mac OS X without additional software to allow us to use the X11 graphics layer. We are still lacking several key features, such as printing, and have issues with saving files. Other features such as the ability to use Java code, Mozilla LDAP support, and play sound during presentations are also broken. There's a whole litany of missing features on our roadmap webpage that need to be implemented before we reach 1.0.
"When 1.0 is released [it's just been released—LSP], we'll have one of the best office suites available for Darwin and Mac OS X. It will be feature complete with OpenOffice.org 1.0 on the other first tier platforms, and it will still look like OpenOffice.org does on other platforms. After that, we want to make the best Mac office suite possible for Mac OS X. I hope we'll have enough developers to begin including the Aqua look and feel, use sheets and drawers, implement collapsible toolbars, and perhaps even integrate with QuickTime and other features of OS X. This will help us to truly deliver a great Mac OS X experience without sacrificing the familiarity and power OpenOffice.org brings from its Windows and Unix heritage."
Dan: "First, the Aqua VCL build needs to actually compile, then we need to fill in the missing implementations. Next, the look and feel needs to be customized to match the Mac OS X user experience and the nuances that Mac users have come to expect. I believe that that's where the most work lies, and that's where the most fun will be had: watching a real office suite take on an Aqua L&F [look and feel] and come out usable and shining, a pleasure to look at and use.
"But not to forget the X11 VCL build either: There are no full office suites available for Darwin, an OpenOffice.org will be the best one when we're done with it. I don't think either of these will happen for months, but we're working toward these goals, having a lot of fun and frustration getting there, and feeling that we're contributing something that will be much used and greatly appreciated."
The Group
LSP: How many of you were working on the project?
Ed: "There were about ten overall contributors to the OS X port since before it was open sourced. All were volunteers from the community and played around with OpenOffice.org in their spare time, mostly looking into how to fix up the errors that made us unable to compile the project. Three developers were active during the few months leading up to our initial milestone release."
Dan: "Like Ed said, there were about ten people in various levels of participation on the project. When it was released, some people took a stab at it, but there wasn't a critical mass of committed enough people to move it forward. I guess they got discouraged and probably put OpenOffice.org aside for a while. Then others came along and began to slowly move forward (and I mean slowly) with the build. Once a couple of people had an idea of what was going on and how to get things going, that attracted more people. Along the way a number of small but significant successes (X11 build, VCL bug in 10.1) contributed hope and the sense that something was happening."
LSP: How did you arrange work?
Ed: "We ended up arranging our work just through contacting each other by e-mail and instant messaging. We were all kind of doing the same thing, fixing build errors, until the final push when we just decided what we wanted to do: come up with installers, write frequently asked questions, etc. With only a few people, it was relatively easy to find hard lines between tasks … one person to author it (or program it), and another to proof it (or test it). When more people join the project, however, it will be essential to use a good collaboration tool such as CollabNet provides. Not only does it provide a good way of disseminating information outside of a core group, it provides historical archives and perspectives. Without the ability to see what had been written and coded between April and September, I wouldn't have been able to get up to speed on the project as quickly as I have."
Dan: "I think the biggest problem that faced the project up until about February 02 was the organizational aspect. There was no one person in charge of the port and so no one had a sense of organization. Everyone felt like they were about the only person working on the thing and that there was so much to do. But now with Ed being the nominal project head, there's more of an idea of how the project is going and what's going on.
"There is no formal to-do or responsibility list. People basically just volunteer to work on something. This of course means there is duplication of effort in some areas, but there are still not that many people working and there's lots to do. Besides, more people looking at a piece of code or a build error means more people with ideas on how to fix it. During the ramp-up to WWDC there were about 4 or 5 people who were part of the core Mac OS X/Darwin team who were working on stuff and a couple of people inside Sun [and CollabNet] who were on the marketing part. It was hectic and a bit unorganized but it was a lot of fun and people just kept stepping up to make sure the job got done. Stuff gets done because people volunteer to do it, not because responsibility gets doled out. And because they volunteer, I think they have more of an active interest in getting it done, and are more likely to do a better job, because they want to. Since there were only a couple of people working, it was fairly easy to organize stuff informally, but this will get harder over time as more people join."
LSP: And how was the group formed?
Ed: "Things formed just from people looking at the websites available on OpenOffice.org and through e-mailing each other on the dev@porting list. There was never any official organizational effort, just a bunch of guys who were interested enough to plod through the very raw and incorrect original material up on the websites, correcting it for each other, and starting to help others get on board quickly. We've continued that style of effort, and hopefully we're making the entry curve for the project a lot easier for newcomers."
Dan: "Like Ed says again, people just got interested, made contact with other people who were also interested, and started asking questions. Those with some answers to give stepped up, and people stuck with it. Those who tried and didn't stick with it probably got discouraged because there was no critical mass of people to work out things with—not enough people who knew what was going on. We hope they come back now: we need them.
"There was also a big discussion earlier this year about whether to fork a separate project for the OS X port or whether to keep it in the porting one. We could have had a separate mailing list just for Mac OS X, but we decided to stick within the porting project since that was easiest and guaranteed easier access to other developers who knew what we did not (i.e., general OpenOffice.org stuff and not just Mac)."
LSP: Towards the end, the group was working tightly and somewhat in secrecy; at least not all of the communications were out in dev@porting. What effect did that have on your group?
Ed: "A lot of issues that we faced came back to a lack of time. Being able to simply IM [instant message] each other when we were available or communicate off-list allowed us to really concentrate on what we needed to do: code. It unfortunately takes time to answer people's questions on mailing lists and help newcomers to the project, and we simply didn't have the time. As more people join us, we'll be able to divide our efforts so someone can spend the time to continually update our website and respond to the community mailing lists. With only two to three people working two hours a day in varying states of sobriety and trying to hit a major deadline, expending any effort away from coding would have not allowed us to move with such speed."
Dan: "As of a couple weeks before WWDC, all the important traffic about the OS X/Darwin build was off-list. It was kind of unfortunate because those few people that did post on dev@porting with questions and who were not on the private list didn't know they were out of date, or would be in a week. We couldn't really answer their questions. Anyway, I think the off-list discussions and communication, while not the best for the port as a whole, did enable the members to get the WWDC build working in a very quick and efficient manner with a minimum of static. Plus, the whole press thing, where we could not afford to have the general Mac community get all worked up over a developer build and then get subsequently let-down when they figured out it actually was a developer build. I think the off-list list also enabled the people involved to get to know each other better and to work better as a team to get what needed to be done."
LSP: How long did it take your group to get to this point?
Ed: "We did not have a reproducible build process until towards the end of march, and a major bug preventing the OpenOffice.org Writer module from launching wasn't fixed until about the middle of April. It's taken since last April to get this far."
LSP: What was the main stumbling block?
Ed: "Technical difficulties present within Mac OS 10.0.x. When Apple released 10.1.x with dual level namespaces, a big roadblock was removed."
LSP: Were there periods when it seemed this effort would fall apart?
Ed: "When I started to see the compiler fail on certain files that worked on gcc on other platforms, I started to think that it might not be possible to make headway until Apple fixed the compilers to not segfault. Thankfully a few clever people were able to find workarounds for the faults in the tools. The next set of development tools from Apple should be a lot better for C++ programs and should solve those problems that we didn't have the expertise to solve easily."
LSP: Where would you suggest people start?
Ed: "I would suggest that everyone start by downloading the pre-alpha milestone binary and start playing with it. :) The amount of time it takes to get up to speed with the initial compile is a while, but I hope that seeing how good the product can be will convince people that the effort is worth it. Also, end users should start there so they can witness what we may be capable of in the future.
"The next step is probably to read the roadmap for the port. It shows you (roughly) where we've been, where we are, and what's left to go. Again, the roadmap is just a reflection of what's been discussed by the community of both Mac OS X and non-OS X porting developers in the past and is always up for commentary or change. Perhaps if we get enough people to help we could even start putting potential dates of completion on it as well …
"After getting a compile started from reading the build instructions, it's probably best to start reading some of the white papers describing the architecture of the program. It's quite large and is difficult to get a grasp on without the documentation. Also, it'll help get people familiar with the terminology and acronyms thrown about on the dev lists and other sources of information.
"There are links to the binary installer at the top of the Mac porting page, and there's a pointer to the white papers at the bottom."
Dan:"I'd suggest they sign up with the Porting project, and subscribe to the dev@porting list, then follow the build instructions; and most important, ask questions when things don't work, so we can help. If they just get discouraged and walk off, that doesn't help anybody. But if we can help people see that things can work, then great. We need the help.
"I'd like to thank everybody who helped of course, but mention Ed specially, because he's the nominal head of the project and someone had to do it. We can all whine to him when things don't work out and ask him to make it better. But then he'll just tell us to deal with it ourselves and report back how we did."
The new version, OpenOffice.org 1.0 for Mac OS X is available now. Thanks Ed, Dan, and the rest of the Porting Project.
