A recent source code release from VIA, maker of graphics controllers popular with Linux users, curiously included a proprietary disclaimer, prompting many to assume the company had reverted to the old, closed license. However, it turned out the addition of the proprietary license was simply a mistake, as VIA Arena Editor and Web Media Liaison Fiona Gatt explained in an email last month.
"The source package released last week was prepared in March 2005, before VIA changed its policy regarding source code package releases," Gatt wrote. "It was an oversight and a mistake that the header was not changed before it was released."
However, to those in the open source community who deal with VIA and the opened drivers -- including Unichrome developers Ivor Hewitt and Luc Verhaegen -- the latest gaffe is indicative of a larger failure in strategy and support, and follows on previous questionable behavior interacting with Linux and open source developers.
Honest mistake, bad mistake, indicative
While the undead proprietary license can be chalked up as an honest mistake, it is a side effect of VIA's lack of a serious open source initiative, Hewitt said in an email.
"What it really shows is that there isn't any real open source development going on here," Hewitt wrote. "Where's the interaction with the community? Where's the release early, release often philosophy? We shouldn't need to be pointing these things out."
Hewitt also criticized VIA's use of the open source drivers in press releases, calling it basically a sham.
"For example, the current 'VIA releases new open source package' announcement is about a chunk of source code that predates their original 'VIA frees source' press release," Hewitt said. "This is not new, it's not news, it's just spin. It is indicative of the fact that at the moment this 'open source initiative' is just a bit of window dressing for marketing purposes."
In terms of what should be expected from VIA going forward, Hewitt thinks not much.
"It would be great if they got their act together and found out what they were doing wrong and what they could do better, but given their track record, I can't see that happening. I'm frequently, openly critical of VIA, and I'm sure I'm not their most favorite person, but I'm just so frustrated by what I see (or rather don't see). They could be doing so much. There are huge opportunities here and they're missing the boat completely."
Hewitt also referenced another ongoing issue dividing open source developers and VIA: copyrights. Hewitt said although the recent publication of the proprietary license with the source code was an honest error, he expressed concern over VIA's apparent removal of license headers from the Direct Rendering Manager (DRM) code, which was replaced with VIA copyrights.
"They have still stripped all of the pre-existing copyrights from all of the DRI/DRM code they've used, and they haven't addressed that issue yet," he said.
In addition to VIA-generated issues, there has been some conflict among the ranks of open source developers working on the Unichrome project and the open source drivers for VIA's products.
Ivor Hewitt said he had left the Unichrome Project, along with Thomas Hellstrom and Andreas Robinson, reportedly after having issues with fellow developer Luc Verhaegen's "confrontational style."
"However, both myself and Thomas are still actively working on the code and producing updated drivers, solving users' problems in the mailing lists, and getting the code into the mainline Xorg source," Hewitt said. "We're progressing nicely with the completely open source drivers, as of the next Xorg release, it will provide support for the Unichrome and Unichrome-Pro boards out of the box. Also providing full functionality MPEG acceleration are Xine, MPlayer, and MythTV."
For his part, Verhaegen said his involvement in the Unichrome project may also be ending.
"I'm not happy with this underhanded stuff, as I'm a pretty head-on sort of person," he said, referring to other Unichrome participants' continued representation of the project, despite having supposedly left.
Typical behavior, taking the bad code
"As far as I can tell, it's a typical VIA cock-up," Verhaegen said of the proprietary license. "Mistakes this bad can't be dishonest. VIA just doesn't seem to be able to get its act together. My guess is that at VIA, there is no one working exclusively on these drivers. It is all very random, uncohesive, sad really."
Verhaegen then described some of his dealings with VIA, indicating the company's clumsy connection to open source developers and code.
"Last time, VIA made a big press release about releasing source which it had been making more or less available before," he wrote. "The actual improvement then, for me, was the fact that they allowed me to change their MIT disclaimer to the OSI-defined one, which disclaims all authors and copyright holders, not just the original author or copyright holder (a common problem). [VIA engineer] Joseph Chan claimed that subsequent versions of their tree would carry that OSI-defined disclaimer. I was quite amused to see that they just dropped MIT on a large amount of files instead."
Verhaegen also detailed his concerns over VIA using code from the open source efforts and removing copyright statements.
"It is clear that VIA is slyly taking a look at what has been happening at unichrome.sf.net," he said. "There are changes which unmistakably were written by me. Given the nature of these specific changes and their size, I'm not going to bother with claiming copyright. I have every right to do so though, as the most striking one is a plain copy of what can be implemented differently. In fact, I'm going to implement it differently, as I'm not happy with it. It has caused some minor problems in the past and there are way more elegant solutions for what it tries to solve. So apparently, even when VIA are taking stuff from us, they manage to pick just those things which are unclean.
"About the removed copyright statements, this is nothing new. They have done this before, but it is up to the respective copyright holders to do something about it. Back then, it was my understanding that these copyright holders were aware of the problem, but didn't bother doing anything about it. I haven't looked into this new claim myself, as this will probably not affect me directly."
Verhaegen indicated questions about the disclaimers from NewsForge may have prompted VIA to make some changes, which yet again fell short of the open source developer's expectations.
"Their recent FB release has suddenly fixed all MIT disclaimers," he said. "The Via Arena announcement of this release is quite telling towards VIA's attitude concerning licensing. Instead of boringly stating that the MIT disclaimer was changed to an OSI standard one on these files, Fiona copied a short part of the MIT license and then cut it short with ' ...etc etc :).' That last bit reads, to me, as, 'but who cares anyway.' It sems that they completely fail to grasp the importance of licensing, even when licenses are as liberal as MIT."
Changes and chances
VIA's Gatt said the company has listened to suggestions from "hobbyists and independent developers," and has made some changes concerning public access to driver source code.
"This decision was not entered into lightly," she wrote. "VIA does understand that the efforts so far do not completely satisfy the open source community. We have, however, received more positive comments and praise from the community than negative."
Although she did not respond to questions about the copyright concerns, Gatt did indicate VIA has a group devoted to the company's open source effort.
"We thank those people in the community who can forgive our mistakes and work with us with a friendly, positive attitude to continue to improve our efforts," she said. "VIA has a dedicated team to support the open source community. Both Windows and open source users are equally important to us."