openSUSE Hobbled By Microsoft Patents

kripkenstein writes "openSUSE 10.2 no longer enables ClearType (which would improve the appearance of fonts). The reason given on the openSUSE mailing list for not enabling it is, 'this feature is covered by several Microsoft patents and should not be activated in any default build of the library.' As reported on and discussed, this matter may be connected to the Microsoft-Novell deal. If so, Novell should have received a license for the Microsoft patents, assuming the deal covered all relevant patents. Does the license therefore extend only to SUSE, but not openSUSE?"

Prior art

[Ph33r th3 g(O)at]

Steve Gibson pointed out decades-old prior art [grc.com] that would invalidate the Cleartype patent (if our patent system weren't corrupt) several years ago.

Novell is the Judas Goat.

[140Mandak262Jamuna]

I think Novell has become an wholly owned subsidiary of MSFT and is being used for the express purpose of setting up precedents and creating more and more FUD. I have seen a version of anti-aliasing and sub-pixel addressing [grc.com] way back when in, of all places, grc.com.

Prior art?

[140Mandak262Jamuna]

I have not been to GRC.com for a long time, I quickly grabbed the URL and posted it here in another thread. Looks like that site cites a long list of prior art. [grc.com] Makes the OpenSUSE's decision even more suspect.

anti-aliasing makes me need glasses

[stokessd]

That might be a good thing(tm). In many cases I prefer non anti-aliased fonts. I have a nice LCD with a DVI connection for a clear picture, then I'm supposed to fuzzy it up? Anti-aliasing lakes me think I need glasses in many cases.

Sheldon

Exactly

[Anonymous Coward]

Cleartype is just sub-pixel AA which existed long before MS ever used it for font rendering. Bytecode type hinting is patented by (IIRC) Apple, it is usually disabled in Freetype and and an alternative (auto-hinting) method used instead.

Apples and oranges, the bug reporter is confused or trolling.

It's only the filtering

[oergiR]

AFAICT, subpixel rendering is not disabled, only the 5-tap filter that's supposed to reduce colour fringes. See http://www.grc.com/cttech.htm [grc.com]. Apparently this is one of the things Microsoft has patented, and I haven't seen any "prior art" for this specific technique. In my humble opinion disabling the filter is not much of a loss as it just makes fonts look fuzzier.

Freetype library is GPL

[mysidia]

As a result, if you hold a license for a patent that is required to redistribute/sell Freetype (or any piece of software covered by the GPL), then, to comply with the GPL you have two options you must EITHER: (1) not distribute the software, OR (2) the patent license must permit anyone's free use

The relevant GPL section is the preamble To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all. , and under Section 7 of the GNU General Public License: For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program.

This means for instance, that Novell would not be free to provide users of SuSE the benefit of a patent license to use a certain feature of a GPL'ed library or software program, and deny that feature to openSuSE users.

Re:It's FreeType for a start!

[segedunum]

That doesn't answer why it was OK to have this enabled before, and has then somehow become a big no-no.

Re:Novell is the Judas Goat.

[duncanmhor]

Judas Goat - used at an abbatoir to lull animals into a false sense of security.

Clear-Type replacement

[Kim0]

They could just use this replacement, which is not patented:
http://oyhus.no/SubLCD.html [oyhus.no]

Re:anti-aliasing makes me need glasses

[xoyoyo]

Depends on the quality of the implementation and the quality/density of the screen. Ironically, the inventor's implementation is poor.

On a 72dpi LCD attached to a PC running Windows the effect is obvious (and hideous) all the glyphs have red and blue fringes. Turning ClearType off is the first thing I do on a Windows box after disabling the Windows XP theme.

On my 100dpi+ MacBook Pro I had to use the zoom function to confirm that it was using sub-pixel anti-aliasing. Even on my second monitor it's acceptable, and that's a cheap low density screen.

Apple have spent some time getting font anti-aliasing right: the initial AA in OS X looked like someone had just applied gaussian blur to the whole screen. Now it actually does what it's supposed to do, which is reduce eye fatigue.

On the other hand, once we get our long promised 300dpi screens monitor resolution will be the same as paper and we can dump kludgy hacks like ClearType.

Correction. AA support in Windows.

[burnttoy]

Windows has had AA text in the following formats.

  Right Click (or Right Menu Key) -> Properties -> Settings Tab -> Tick "Smooth Edges of Screen Fonts".

WinXP - ClearType fonts supported (at least on Pro) - get a control panel applet from msdn/microsoft.com to change settings. HW support via alpha blending.

WinXP Tablet Edition - Support of 90 degree rotation e.g. aliasing in Y instead of X (screens mounted portrait)... I think I'm right on this.

Vista - more of the same I guess!

YMMV - It's been a while since I mucked with Windows GDI Drivers.

Re:never so

[oergiR]

Mod parent up.

The main developer of FreeType decided to disable the filter [mail-archive.com] in September. The Novell deal was later and had nothing to do with this.

Re:Prior art

[pikine]

I'm afraid the decades-old Apple II and IBM PC is not prior art. Pixels are either on or off for Apple II and IBM PC's CGA displays, so they apparently don't (and can't) care too much about color fringing. Sub-pixel font rendering on LCD screen deals with 256 shades for each sub-pixel, and the emphasis is on how to adjust sub-pixel brightness to reduce color fringing.

This is explained in Steve Gibson's Turning Theory into Practice [grc.com]. Sub-pixel font rendering is not the same as sub-pixels on CGA displays. The ideas are related, but the plumbing is different.

Perhaps I'm misleading in saying that CGA is not prior art of ClearType. I haven't actually read the patents of ClearType, so I obviously cannot tell; I'm basing my claim solely on Steve's webpage alone.

Re:It's FreeType for a start!

[Movi]

Yes, indeed Cleartype sucks, and ive not even known you can have it under linux. For an ever better font setup you can enable BCI in freetype and have freetype display font quality on par with Mac OS X (which nobody can dispute displays the best quality). For example ubuntu people can download debs with prepatched freetype here http://ubuntuforums.org/showthread.php?t=343670&hi ghlight=feisty+fonts [ubuntuforums.org].
The standalone patches are here http://david.freetype.org/lcd/ [freetype.org]

Re:Suse vs Open Suse

[MollyB]

I just last week switched from SuSE 9.1 to Ubuntu 6.06 (Dapper Drake) on my laptop. The old OS always crashed or froze, and was a tempermental beast that chronically corrupted the Reiser FS. The Ubuntu install was less than 20 minutes (although downloading of package upgrades took an hour on DSL) and has been running superbly 24/7 since boottime. I hope you are happy making the change; I most certainly am!

BTW, if you are a Windows person who is looking for a friendly Linux distro, this is for you. You can run Ubuntu from the CD to try it out before installing, and it is the very epitome of "user-friendly." You don't even have to edit config files, if that seems daunting. Take the plunge--you won't regret it.

Re:Prior art

[Anonymous Coward]

Prior art doesn't have to be identical - patents are supposed to be for something novel and nonobvious*, not for a logical progression given technological development. i.e. the fact display subpixels didn't have 256 brightness levels back in the day doesn't mean that it isn't blitheringly obvious that IF THEY ONE DAY DID, similar techniques would naturalyl apply.

* Yes, in practice the USA grants patents for basically anything regardless of merit, but the USA sucks.

Re:Prior art

[spitzak]

Holy confusion, batman!

That was completely misleading. The use of 256 levels for making antialiased fonts is really old and has nothing to do with this.

What sub pixel rendering does is make the pixel represent *more* than 256 different possible combinations of the fg and bg colors, where combinations are how an fully opaque edge falls into the square the pixel represents. Exactly how many is unclear, it is not 256^3, but I think it is 3*256 for the case of an antialiased vertical straight edge of an object significantly larger than a pixel (it goes down as the edge gets less vertical or if another edge gets within a few pixels of it). This is done by using hardware quirks so that various of the 256^3 possible colors are interpreted by the viewer as different coverages.

The Apple technique made a pixel represent more than the 2 possible coverages you mentioned. This is done by using hardware quirks so that the some of the 4 (16? i dunno) possible colors are interpreted by the viewer as different coverages.

They certainly are related, but I don't believe enough to invalidate Microsoft's patent. The difference is that the "hardware quirks" are vastly different.

Re:Prior art

[Sleepy]

I remember using anti-aliased fonts in "Spectrum 512", a terrific graphics package for the Atari ST series computer. With it, you could optimize images for TV. If you could not afford an Amiga, this was a very impressive way of generating titles and graphics for video. The Atari ST had a composite NTSC output port and the horsepower to drive it.

I remember the same technique used on the Atari 8-bit computers, in the monochrome "Graphics Mode 8" level. By offsetting the *placement* of pixels, you could accomplish new colors. The effect was used (but less noticable) in lower-res modes, which also had more color, but you could choose between "smoothing" OR "more colors".

Note that these were COMMON techniques on computers that hooked to TVs, and also to CGI monitors. It was especially useful on the Apple 2c computer which had a monochrome high res mode like the Atari 8-bit.

I see a lot of posts here confusing these old techniques with "anti-aliasing". This is NOT the technique used -- anti-aliasing simply blends the difference between 2+ contrasting pixels. These old techniques did not do that - they CALCULATED the placement of pixels to take advantage of "display artifacts" - and generate perceived resolutuins (or color range) higher than the hardware was intended to deliver.

It would be difficult for someone else to argue Apple's prior art if Apple is un-interested in the battle, so this patent is not going anywhere.

Re:Prior art

[someone300]

I just looked at a screenshot of "Cleartype" on OS X (From http://www.microsoft.com/typography/ClearTypeFAQ.m spx [microsoft.com]). It looks revolting. I use Linux and OS X with subpixel rendering enabled. The first three lines look OK, but it starts to look blurry after that.

Screenshot of my OS X system: http://img248.imageshack.us/my.php?image=picture3r p7.png [imageshack.us]

If you zoom in, you can see I'm definitely using subpixel rendering. I get the odd blurry looking font on my OS X system, but nowhere near the sort of stuff I see on other people's Windows laptops.

Re:Prior art

[AJWM]

The Apple II didn't have a CGA. Colors were generated by dot-timing the luminance signal into an NTSC composite monitor (read, TV), faking out the color decoding in the monitor. Subpixel rendering was done by reversing that to choose the appropriate color to generate the desired dot timing. On a monochrome monitor these showed up as higher-resolution dots than the nominal pixels in display memory.

Go look at the circuit diagram for an Apple II, for pete's sake. It's not that complicated, maybe a dozen or so 74-series chips plus the memory and CPU.

Clear type uses exactly the same idea -- pick the color to activate the desired combination of R, G and/or B stripes in the LCD pixel -- i.e. activate the desired sequence of horizontal dots by color choice.

Licenses.

[miguel]

Just to clarify the thesis of the post.

Novell has not received any licenses to any patents, and neither has SUSE, nor OpenSUSE.

The Microsoft-Novell agreement is about not suing customers over any potential patent infringement.

Since OpenSUSE is a community effort, and it is used by people that might not be customers of Novell, removing code that is known to infringe on a patent is the correct thing to do (same policy applies to Mono).

Re:It is about precedents

[drinkypoo]

Today, you can only make a have baked OS and have half baked applications before you enter the patent and copyright mine field.

This realization is precisely what the patentmongers fear. Because if this realization gets out in the world it will do them irreparable damage - the argument [in court] will be that it is utterly impossible to develop software because of the patent system. Admitting that you have a problem is the first step on the road to recovery...

In my opinion, open/free Linux is still 5-10 years behind Windows and OSX in terms of desktop functionality and these hurdles aren't going to allow the gap to close any time soon.

Your opinion isn't worth much. There is one place, and one place only, where Linux is behind. That area is in drivers. Oh sure, Linux might support more hardware than any version of Windows - I wouldn't have any trouble believing that. But the problem with Linux is that the support for some hardware raises numerous problems which you simply don't have to deal with on Windows.

Aside from that, can you name one kind of task that's easier on Windows than, say, on Ubuntu Feisty? I really will settle for just one clear example.

Very Similar != Same in court eyes

[Tablizer]

The courts seem to favor the patent defendant (such as MS here) when issues of "obviousness" come to light. It seems it has to almost be exactly the same thing for a judge to overturn it, not merely very similar. A similar issue appeared in the GM (or was it Ford?) break-pedal case. Obviousness was pretty well demonstrated, but it was not deemed obvious *enough*. However, that case is still on appeal and climbing its way up the courts and may affect this case also because the level of obviousness is the key issue. Apple II using a very similar technique may not be enough. It may have to be pretty much exactly the same thing with the exact same look to qualify. (I think this is stupid, but that is another issue.)