This is our 6th interview with Wine developers. Check out the Interviews page for previous ones.
This week Andreas Mohr finds himself in the hotseat. Andi was born in Karlsruhe, Germany in 1977 and grew up in Renningen, near Stuttgart. He did the usual military service after high school and in 1997 began studying electrical engineering at Stuttgart University. Now he's attending the University of Applied Sciences in Esslingen studying computer science. Besides the normal CS classes Andi is focusing on embedded systems, automation, and networking.
BV: How did you get involved with Wine?
Andi: Well, it just happened that someone told me that there was that cool OS that I just had to try, so I tried it and soon found my way to Wine, since I was interested in also running some Windows programs on Linux. And I believe everyone would agree with me that Wine at that time (1996) was so incredibly crappy that I just had to do some major contributions to it ;-)
Besides, I had an extensive DOS Turbo Pascal and Assembler lowlevel programming background, so I was foolish enough to "fall in love immediately" ;)
BV: Do you remember the first patch you submitted?
Andi: Of course I could look at documentation/ChangeLog.OLD, since I'm not entirely sure right now, but from numerous previous glances at that file, I'm pretty sure that it was a patch fixing some NE file (Win16 programs) data segment size calculation, way back at the end of 1996 or so.
BV: Did Wine even support PE executables back then? Did you have any hand in getting the loader to work with them?
Andi: Yes, as far as I remember there was basic Win32 executable loading support at that time. But the real Win32 support started later. I did improve PE support on several occasions, but I don't remember any details (see ChangeLog for info).
BV: You've been to a lot of conferences and given quite a few talks on Wine, do you find it helpful to meet other developers?
Andi: Most definitely! For example, doing an internship at CodeWeavers has been VERY interesting and entertaining (boy, did we have bloody ping pong matches!). And I really can't say that the wineconf 2002 conference, graciously sponsored by Lindows, has bored me. wineconf 2002 has been the first time that almost the entire Wine developer community (over 30 people!) managed to gather somewhere and talk about pretty important topics. We had a lot of fun, that's for sure!
BV: Who did you really enjoy meeting at WineConf?
Andi: Hmm, I'd say Gavriel State from TransGaming was certainly interesting. Also, I'd never met Ove Kaaven before (even though I'd been chatting with him on IRC for an amount of time which could almost be measured in months). And in fact ALL of the people meeting at that conference were very interesting to talk to.
I loved talking to Jeff Tranter (formerly Corel) to get some background information about the huge help Corel was for Wine when they used Wine to port their programs. We had some nice beers one evening :))
BV: At the time of WineConf the license had just changed to LGPL. Was that a popular topic of discussion? Or was everyone over it by that point?
Andi: Ah, no, far from it! While the rest of us were doing a "normal" discussion, there's even been a second separate "licensing" discussion going on between CodeWeavers and TransGaming. They were forbidden to bring any weapons to that discussion, though! *grin* (the discussion has still been relatively "silent", as far as I know)
BV: Is Wine just a hobby or have you gotten paid to work on it?
Andi: Well, I'd say it's mostly a hobby. But I've been paid to do Real Work on some Wine parts by various companies on several occasions already.
BV: What are you studying right now?
Andi: Yep, I'm studying the worst subject you could possibly be studying right now, given the current economy: Computer Science ;-) When I started doing that, people used to say stuff like "whoa, you're studying CS?? you'll surely be able to get a great job and earn a 'certain' load of money once you graduate!". And frankly spoken, I can clearly remember that I was sometimes opposing them at that early time already ("there'll surely be a time when this will prove to be dead wrong, and hopefully it won't affect me too much yet once I'm finished").
Anyway, since I'm almost finished, it'll just take one additional year until I graduate. And I really hope that the situation will be somewhat less catastrophic at that time... ("it really can't get any worse than that..." -- or can it??)
BV: What's your favorite part of Wine to work on?
Andi: Hmm, maybe installers or file system layer. And maybe the program loader. But apart from that, I just work on about everything that needs fixing and that interests me.
BV: You've probably written the most documentation of anyone. Do you like working on that, or do you do it because no one else is?
Andi: Hmm, sometimes I like doing that, but in many cases I don't. And I can't say that I've had a whole lot of support or suggestions so far, not even after having added a special "comments" section to the Wine User Guide... So in general, feedback or help is very poor.
Also, I don't like that there are some developers that submit enduser relevant code without documenting a single part of it (yes, in many cases I can easily find out about its functionality, but having to do so grows tiresome rather fast, believe me...).
Oh, and by the way: I'm currently planning to submit the most extensive Wine Users Guide I've ever done pretty soon. (hint, hint! ;)
BV: Since I initially asked that, you've submitted some of the update you mentioned, Dimi made some organization changes, and we've also had WineHQ revamped. What areas of the docs need work now?
Andi: Hmm, I don't quite know. Just look at http://www.winehq.org/?page=todo_lists , then you can see the current status. I know that the Wine User Guide could need a lot more work, but of course this is always the case (documentation is never finished).
BV: You've worked on a lot of Winelib utilities. Do you see that as one of the last major areas that could be improved? Or are there still core features to add?
Andi: Well, Wine definitely needs more auxiliary utilities, such as a real graphical control panel (no, programs/control/ is not enough yet, since it only runs one particular panel application!) or a desktop or the wineboot utility. Unfortunately I haven't been able to do a whole lot of wineboot work recently due to "other" very time-consuming tasks (see above ;), so Shachar Shemesh stepped in and submitted the first parts of it (mucho thanks!).
BV: Alexandre said your wineboot patch was overdesigned. I didn't really examine it, but what was the problem?
Andi: Hmm, I probably used too many defines, and I "reused" some functions too many times (since the various bootup parts are doing very similar things). And I guess he didn't like the amount of configuration settings I included. (although I think that that's needed, in order to avoid becoming the next popular target for virii and trojans). Add to that the fact that I've been very busy during the last semester (and also the fact that I've been writing an ALSA Aztech PCI 168 sound driver), which caused proper wineboot maintenance to be a problem, and then you know why it's been less than optimal. I plan to eventually do some wineboot improvements soon, though (since I've got some free time soon). But then don't hold your breath, since my ToDo list is looooong.
BV: Why did you decide to write a sound driver? Have you worked on any other parts of the kernel?
Andi: Well, I discovered that Aztech PCI168 card in a PC of my father, and I found out that there were zero, zilch Linux drivers for it, so I kept that in mind for quite some time. And about one year later I started implementing a driver, from scratch.
I've written the NI5010 network card driver (you will not even remember it, since it's an ISA 8bit (!) 2MBit (!) ethernet card from stoneage), together with Jan-Pascal van Best, who was writing a driver for it at the same time, so we combined our fortunately rather diverse efforts. That was back in '97.
BV: How does writing a kernel driver compare to working on Wine?
Andi: Well, writing a kernel driver is somewhat "strange", since it's unknown territory compared to the pretty familiar Wine environment. But apart from that I'd say it's comparable. Both projects have a huge source tree and you need to find your way around in both.
Trying to get your work accepted into the kernel can be a bitch at times. For example, I discovered a complete deadlock in the 2.4.x kernel when using SB16 in full-duplex mode (these were actually two separate fatal bugs!), and I sent a mail containing an explanation and a possible, working patch to the Linux kernel mailing list, but I've never had any real success in getting that fix in. The reason that nobody cared about it is probably that 2.6 will use ALSA instead of OSS sound drivers anyway... But in other cases I've managed to get patches accepted much faster. E.g. I recently fixed a software suspend locking bug in 2.5.x.
BV: When you finish your driver work are you planning on working on Wine more or do you have some other big projects planned?
Andi: I'm afraid that I have to tell you that I started yet another project: providing an Open Source Linux driver for the shamelessly neglected ACX100 22/44Mbps wireless chipset by Texas Instruments, at http://acx100.sf.net .
The various inofficial binary driver versions "leaked" onto the net are pretty insufficient: they only work with Linux 2.4.18 specifically, and they're more or less buggy. So far we're making good progress: we now have a driver which initializes the card and handles ioctl()s. 4 out of 13 developers are currently working on the code, so we have a sufficiently large amount of manpower. I expect to have a preliminary working version in about 3 weeks or so (don't hold your breath!).
BV: If there's one thing you could implement in Wine, what would it be?
Andi: I guess what would be useful is some kind of Wine capability to do automatic in-memory patching of programs in case they are "broken", i.e. they abuse some horrible Windows "feature" and there's no way that Wine can properly implement that. So I'd let Wine have a database comparing the program it loads, and if it finds a match, then it would apply whatever binary patch/hack is needed to make that program run on Wine despite its huge problems or bugs.
Of course technically I could implement it, but it's always a matter of time... And to tell you the truth, other things are way more important currently (such as further docu work).
Not to mention that implementing such a feature could actually conflict with "cool" things like the DMCA...
BV: It wouldn't be too much different than a copy protection crack. But isn't Wine supposed to implement "bug for bug" compatibility with Windows? Are there any particular braindamaged programs you've used?
Andi: Well, "bug for bug" compatibility in rare cases simply ISN'T possible (take VxDs or direct descriptor table access as an example). Or other nasty things might include programs accessing Win16 GDI objects in a way incompatible with "modern" Windows versions...
I don't remember any particular braindamaged programs, but this isn't to say that there aren't any - you can easily find several if you search a bit.
But all in all I keep being surprised at how well we're able to reimplement the Windows mechanisms... (surprisingly few real issues for such a large undertaking with such a different operating system).
BV: You mentioned the DMCA, I've always thought that was a good example of people not caring about something until it was too late. What do you think about some other digital rights management initiatives being proposed?
Andi: TCPA/Palladium/NGSCB/DRM "Trustworthy Computing" initiative is a big threat to the IT world as a whole as we know it (and love it), and especially to "non-conforming", "improper", "cancerous" and "un-American" software such as Open Source software and freely shared software.
If you're not yet aware of these possible restrictions of your Personal Freedom (to explain it in simple words: you will not be able to use your PC the way you're doing now, since it'll tell you what to use or not to use and how or how not to use it), then feel free to read about it at:
Note that the second article talks about "it needs to stay out of the consumer space for a while". Translation: "we know that we have some severe issues with enlightened and grumpy endusers, so let us shove it down by introducing it in the commercial space first, which is quite eager to protect company secrets anyway. Once it's successfully introduced, resistance by endusers will not persist, since NGSCB will already be too widely used and they won't be able to evade it."
In other words: we'll have to think about a globally organized collaborative effort to boycott this threatening development and to make sure that even the uninformed masses do take notice.
BV: I agree - waiting until someone is shipping it is too late. Thanks for the interview!