I attended LinuxWorld in New York this year, so I thought I'd scope out any printing-related information (in between scoping for actual work-related details, of course). I also record a few other nifty things at the show...
I dropped by HP's printing demo Wednesday, and had a chat with John Oleinik, an HP fellow who gave me some actual info on HP's inkjet driver plans. Keith Burres of VA also said hello; he's a manager of the related work there.
In short, they've indeed got real plans. Come Spring, they will be shipping and offering tentative support for a Ghostscript-based driver that drives the bulk of their modern DeskJets. Details:
- The driver is not DLF-free; the license is a with-source affair plus a "for use with HP printers only" clause. It's otherwise BSD-ish. I asked Bruce Perens about this, and he says that it'll be a few months: the only reason it's not a proper free license is that HP must complete a patent search so they can satisfy themselves that they aren't treading on anyone's patents. They absolutely plan on fixing the license in the future.
- The driver is derived from an embedded (ie set-top, kiosk, etc) developer SDK HP has shipped for a while. Basically they've made that operate on Linux with Ghostscript as a front-end. In theory, arbitrary non-Ghostscript raster-generating software could use the drivers directly, but this would be a special-purpose situation. Postscript and PDF are clearly the way to do general printing.
- HP will offer end-user support, via email only; they will be releasing this first through VA's SourceForge project and crossing their fingers that most users go there for support and help themselves.
- I'm not sure if this will show up in the box, or if penguins will appear on HP packaging.
- The driver does include full support for decent output: most or all of the various PhotoRET and such modes are supported. The code will be there for all to see--color code and all. At LinuxWorld, they were demoing the driver; it did indeed make pretty pictures. I printed lots of samples on the DeskJet 990 they had there, most of which I will be delivering to the gimp-print lead, since he wasn't able to make it.
- The driver does not support any PPA DeskJets; supposedly there is some sort of legal tangle with Microsoft. In theory, the free pnm2ppa project could make use of the code, since it would be for HP inkjets, but then their code would have to be nonfree, and that seems unlikely. Perhaps later, when the license is fixed.
- The driver supports few if any printer-specific features; ink levels, alignment, duplex, banner paper, etc aren't done. The usual "driver" features from the Windows world like poster and email printing are also not offered, but they don't belong in a driver, anyway.
- The driver is in two parts: minor-sounding mods to Ghostscript to support a shared-memory coprocess (as opposed to pipe) sort of "filter", plus of course the "filter" itself. Reputedly the "filter" in fact operates as a persistent daemon, but I got the feeling that Ghostscript still handles the output. I guess they bounce raster and color ops through the daemon from the Ghostscript-end driver (arguably this is an end-run around the GPL, but Ghostscript licensing is already pretty awkward, and the intent is clearly not obfuscation, since the daemon code is readable. I'm not overly concerned yet...).
It's clear that there are various projects that can use the code as is; Mike of CUPS could probably adopt the drivers into native cups ones, and various distributions will probably ship the gnulpr/gpr/rhs-printfilters kit that VA will undoubtedly whip up for HP.
More information can (or will eventually) be found at HP's SourceForge page.
In the meantime, you can discuss or ask questions in linuxprinting.hp.general, and I'll try to get the HP guys to answer.
I bumped into IBM's Mark VanDerWiele; he's part of their OMNI printer driver project. They're making steady progress; evidently the plug-in structure for dithers and such has been tidied up a bit, and most interestingly, they've just added Canon inkjet support.
I chatted with them a bit about this. Their Canon support was written using the NDA information they've had access to over the years. They obtained Canon's blessing to release the resulting drivers under the LGPL. Evidently the story is that Canon isn't yet willing to step up an explicitly do this itself, but this roundabout method of "documenting" their devices through IBM's project was workable.
IBM plans to move OMNI over to SourceForge, since it's sort of languishing somewhere in IBM's DeveloperWorks website.
[End of printing-related events.]
IBM was also demonstrating their Linux Watches. The prototype that everyone has seen on the web was there, along with a more interesting second-generation prototype sporting a really glorious new screen. Instead of a normal glass LCD panel, this watch featured a 640x480 bright yellow organic polymer display. Apparently it consumes only slightly more power than the unbacklight LCD, but it's so bright and sharp that it's like night and day.
I want a Pilot with that display...
Axel had a demonstration of their VNC thin client. It's not quite perfect yet, but it represents an sensible way to get a low-cost terminal. Existing "thin clients" like Wyse or NeoWare's terminals generally contain full-fledged Linux or BSD firmware, an X server, and so forth; they therefore tend to cost more than an equivalent computer. By implementing the VNC protocol directly, a terminal could, in theory, offer full graphical connectivity in a very small footprint. Axel has done this; the terminal they sell has only 1MB RAM. It supposedly goes for a smidge under $300. Alas, the current model only does 800x600x8; they'll have to double that in every dimension first, and cut the cost a bit. Even so, it's sold for nearly half what traditional X terminals usually go for.
I asked if they were planning a wireless webpad-style device, since this is obviously the way to do that, and it hadn't occured to them. Pity.
Thursday evening, I wandered into HP's reception at the Marriot, intending to meet Bruce Perens. While I missed Bruce, I did bump into Greg and Keith, who have been working together on the HP printing stuff. They offered to sneak me into the Revolution OS screening, which had "sold out" moments after popping up on Slashdot. So I got to see it.
Unfortunately, the theater was little more than half full; evidently the Slashdot crowd got lost on the way to the theater. The film itself was quite good; it was the director's first real film effort, financed on credit cards and shot on 35mm film. The fellow was apparently in student government with Larry Augustin at Stanford, so in part it tracked VA's rise, as well as the rise of Linux and free software in general. There were substantial interviews with RMS, Linus, ESR, Bruce, and Larry Augustin, as well as coverage of the FSF formation, Linux history, Cygnus, VA, etc; right up through the Netscape Mozilla release and VA IPO. Even Windows Refund Day was covered.
All in all it was a vastly better treatment than AntiTrust, which treated free software so poorly that I felt obligated to post this article on Advogato. Revolution OS did a vastly better job of capturing the spirit of free software and the worldwide community around it.