LWN talks to CodeWeavers Chief Jeremy White

We have about 20 people, about 12 here in St. Paul, and about 8 scattered about the world, ranging from California to Siberia. It's actually kind of fun - we use IRC for internal communication - and I can chat with employees right before I got to bed ("good morning, Mike"), and then again after I wake up ("good night, Mike") <grin>.

LWN: On April 6th, CodeWeavers cut ties with Lindows. What were the original goals of that partnership? What was CodeWeavers looking for in such a partnership?

It has long been my mission to make Linux into a truly Windows compatible operating system; helping Lindows accomplish that was very exciting to me. Lindows, on the flip side, was looking to inject some energy into Wine and take it to a higher level, so that they could have an exciting message for their market.

LWN: What did Lindows lack that you hope other companies can provide?

Well, if I had a wishing wand, I'd wish that Lindows had as strong a commitment to Wine, and to a Free Wine, as I do <grin>. I can't fault them for flagging in their support - they provided a great boost to Wine as it was.

We are looking to partner with other companies that can provide the same solution - a seamlessly Windows compatible Linux system. The vision is simple: someday, I want you to be able to go to a CompUSA store, buy any piece of software on the shelf, take it home, pop in the CD, and have it automatically install and run as simply as it was designed to run (except of course, you'll be running Linux <grin>).

At this point, the most important thing is for Linux to gain momentum as a desktop solution. So, the most important thing for me is for other Linux companies to help grow the overall Linux desktop market.

LWN: What do you think about OpenOffice.org and AbiWord?

I personally love AbiWord; it's probably my favorite word processor. (But no, I don't use it; I use MS Word, because I believe strongly in 'eating the dog food'). I also think that OpenOffice is great, and I firmly believe that it is a Good Thing (TM) for the world. We need word processing to be based on open standards, not proprietary ones.

LWN: Will a good Office clone remove the need for CodeWeavers' product?

No. I learned a long time ago when I was helping my mother switch from an obscure desktop processor to Word: she didn't want a useful word processor - she wanted *her* program. Our goal is to make the transition to Linux less painful. If you can shift to a Linux desktop and retain your comfortable MS Office applications, then later you can explore OpenOffice and other solutions.

An organization can switch 1000 desktops, 950 of them to OpenOffice, if they know that the 50 people with complex Excel macros will be supported. Later, those 50 folks can be shifted to other solutions.

<soapbox>

It drives me nuts that a vocal minority of the Linux community cannot understand this point. It does not hurt GnuCash to have Quicken run on Wine, nor does it hurt OpenOffice, in my opinion. If we make it possible for another 1 million users to use Linux, then GnuCash has another million potential converts. Further, the GnuCash guys can convert the folks that only use Excel on Linux, and the OpenOffice guys can convert the folks that really only use Quicken.

Without Quicken and MS Office, those million users will never convert to Linux.

It's all about choice - the more choices users have, the better. If someone takes the position that without Quicken they can't use an OS, then *the only way* they will switch to Linux is through our efforts.

Further, by announcing support for Notes, we caused a huge tidal wave of customer queries and outcry at Lotus. I know for a fact that the senior management of Lotus takes the Linux desktop very seriously entirely because of Wine.

Now, 10 years from now, when the Linux desktop is king, and all applications come out first for Linux, *then* I may need to find something new to do <grin>.

LWN: What else, besides CrossOver Office, are you working on?

We're about to announce a version of CrossOver Office which will target the thin client and Unix workstation market. We intend to compete directly with the Citrix/Microsoft Terminal Server solution set.

[ Editor's note: This announcement occurred last Tuesday. ]

LWN: What do you foresee will be CodeWeavers' champion product in 5 years from now?

I think in 5 years, CrossOver will still be our flagship, but I'm hopeful that we will have a set of tools for managing thin client environments that will be our rising start.

LWN: What exactly is CodeWeavers' role in the Wine group?

Well, I like to say that we're the leading corporate supporter of the Wine project. We're pretty intertwined throughout the Wine project. For example, Alexandre Julliard, the maintainer of Wine, along with many of the top Wine developers, work with us. Because we all come from the Wine project, we all care passionately about the success and the integrity of Wine. We also host the web site and the application database. We've put a lot of energy in to the Wine infrastructure - we built the appdb, set up bugzilla, did a complete web site overhaul and wrote a bunch of documentation that no one else wanted to do.

Sometimes this gets a bit tricky; we really don't want to be seen as 'dominating' or 'forcing' the Wine project; instead, we want to be seen as the biggest supporter of Wine. This got a bit tense during the license change (I and Alexandre led a drive to successfully switch Wine to the LGPL). There are still people who feel that we forced this change, with the evil intent of harming Transgaming. Yet, I feel that Alexandre was more than circumspect on this issue. He and all of the CodeWeavers developers felt that a switch to LGPL was imperative, and yet he did not make the switch until he felt a majority of the non CodeWeavers, non Transgaming Wine developers were in favor of the change as well.

LWN: How many full time Wine developers (if any?) does CodeWeavers employ?

We employ about 12 full time Wine developers.

LWN: When will Wine be ready to run almost every single application?

Much too long for my taste <grin>. We're probably still another 12 to 18 months out from 'perfection'. However, I think that Wine is going to start picking up steam rapidly as the year progresses, and I think Wine is going to be doing exciting things on that 18 month journey.

LWN: You mentioned in another story that there are 15,000 function calls in Windows, and that you write one per day. What is the progress on that?

Hmm. I probably mischaracterized the situation. I think what I tried to show was how hard the task was - that if we wrote one per day, it'd take us forever to finish. For the record, there are around 8,000 functions done in Wine. The good, and important, news is that most applications only every use about 500-1000 API calls, and we've got those covered. (It's kind of like English - the OED is huge, but most people only use about 1,000 words on a regular basis).

LWN: How do you prioritize what be written?

Well, we have this dart board...<grin> Seriously, we listen very carefully to our customers. We're hoping to add a 'voting' ability for our customers to more directly prioritize us. We then also weigh how close an application is to running, and make the decision to support it based on those two factors.

However, we have a few priorities that stand above all of the others. First and foremost, we look to solve problems and improve the support for our already supported applications. This gets a bit tricky, because solving Wine problems is very hard, and often you suffer from the asymptopic effort problem: the work required to close the last 10 bugs is more than the work required to close the first 90 bugs. Second, we try to provide a stable and reliable platform, so regressions are the number one evil around here.

LWN: Can you give examples of some challenging functions you've had to write?

Where to start? <grin> Okay, I've already written a novel length answer here, so I'll just give you my favorite undocumented feature we had to deal with. It turns out that when Microsoft Office installs its icons on the desktop, it uses a really cool feature - the icons don't point to executables, but to a fancy COM based object. If, for example, you didn't install Excel, you still get an Excel icon. Then when you click on Excel, the fancy object can go out and install Excel off the CD.

Of course, this fancy interface is *completely* undocumented, so you won't see any other application using it. And, it was great fun to bypass that problem.