Prior Art?

Posted Jan 11, 2006 22:49 UTC (Wed) by filker0 (guest, #31278)
Parent article: Microsoft's file system patent upheld (News.com)

Back in the late 1980s, I worked for a company that produced (among other things) a Unix like OS (Coherent) and C compilers/cross-compilers for various systems (MS-DOS, TOS, Coherent, ...). We supported long symbol names in object files in much the same way that VFAT supports long file names -- there was a short entry in the symbol table that contained a short (truncated/mangled) name and all of the information that was associated with the symbol followed by another entry that contained the long name and a different type that associated the long name with the previous (short) entry. Tools that supported the long names knew how to interpret the extended entries, those that did not handle long names ignored the extended entries and simply used the short names.

I don't know who came up with the scheme, but it was well established before 1986 (when I went to work at MWC). In many ways, this is pretty much the same thing that MS has done in VFAT, using dual entries, one backward compatible with the actual file pointer info, the other with the extended name. MS was certainly aware of this technique, as MWC was a leading vendor of C compilers for MS-DOS for many years before MS entered that market with Quick-C.

Even if this is not exactly the same technique (due to the name mangling for the short names), it's pretty darn close.

Prior Art?

Posted Jan 12, 2006 1:24 UTC (Thu) by neoprene (guest, #8520) [Link]

And you thought M$ never had an original idea?
Prior Art to these lazy monkeys means [searchable] Patents.
If the Prior Art is not commonly known, and well published, like in the form of a Patent, these morons want to understand it to mean that the idea does not [did not] exist.

Prior Art?

Posted Jan 12, 2006 4:41 UTC (Thu) by patmartini (guest, #35146) [Link] (3 responses)

Hello. I too was in the industry then, actually since the 70's. I
remember Coherent. Prior to that I worked with Digital Research, the CP/M
people, and was involved back then (in the pre-Windows era) with an
extended file system format. In fact, the extended architecture actually
began there (migrating into CP/M 86 [later MS-DOS 1.1]). The vfat system
was an extension to the original FAT (later known as FAT16) system,
introduced by MS with W95. This still used 16 bit addressing, capable of
up to 64 megabytes. W95 Second Edition brought FAT32, a further extension
of the vfat system which actually uses 28 address bits (can reach
gigabytes). The other 4 bits were reserved for the long filename support.
This is actually what is still in use today.

This is the crux of the patent point. Gates and company only do things
with dollar signs, and thus, the big targets for this are the memory and
USB-based manufacturers. Flash memory larger than 64 megabytes obviously
require the 28 bit addressing. The long file name is a non-essential
bonus. The secondary target of any patent infringement litigation from
Microsoft will be the Unix/Linux guys, who incorporate FAT32 addressing
read/write capabilities in their operating systems. This obviously forms
a common bridge (a common directory structure) which Gates would like to
see go away. I wonder if this has something to do with Apple's planned
Unix departure.

I know personally that this is a thorn in the Redmond side - dual boot.
They will attack anyone who treads on their preiminent dominance. There
have alreay been several NTFS lawsuits (mainly concerned with writing)
which were the result of reverse engineering. This is why we do not see
NTFS write capability in Unix/Linux. I know Gates, and he will stop at
nothing to satisfy his 'rule the world' attitude (remember Netscape?).

An aside here, I was really upset with them (MS) over the Netscape thing,
but didn't feel as bad when I heard that AOL bought all Netscape assets
for a cool 4.5 Billion (Andreessen and Clark cashed-in big time)! While
the world suffered at the hands of underhanded and illegal MS business
practices, the Netscape guys reaped a huge payoff, thanks to Gates.

I don't think there will be any secondary winners this time, however, as
a shutdown on a common 28 bit addressing scheme will very seriously
impact the industry. I agree with a former comment here about a new,
unified, open standard, but I don't think you will see this real soon on
a Windows box. It's too bad the rest of the industry is so splintered and
separate - a united effort would put some serious (needed) brakes on
Gates & co.

Prior Art?

Posted Jan 12, 2006 16:21 UTC (Thu) by Soruk (guest, #2722) [Link]

Why should flash memory > 64MB require FAT32? We had hard discs far bigger than that long before Win95OSR2 came out. Indeed my camera uses FAT16 on 128MB flash cards.

How is FAT32 patentable?

Posted Jan 12, 2006 23:37 UTC (Thu) by leonbrooks (guest, #1494) [Link]

If it's just the long-filenames part which presents a problem, then that's not a big issue. It should be relatively easy to produce a stupid-patent-compliant large-FAT driver for Linux.

I'm still not clear on how VFAT could possibly be patentable, though. The techniques used existed in VFAT are just riddled with prior art.

Prior Art?

Posted Jan 13, 2006 5:32 UTC (Fri) by khim (subscriber, #9252) [Link]

Default Zaurus SL-C860 kernel does not support FAT32. Yet I'm able to use 4GB flash formatted as FAT16 there - full size no paritions and/or other tricks.

Now I know that my 4GB card is smaller then 64MB... Hmmm... Strange arithmetic...

Prior Art?

Posted Jan 13, 2006 15:59 UTC (Fri) by wookey (guest, #5501) [Link] (1 responses)

I haven't actually read the patents in detail, and IANAL, but the problem here is the 'not exactly the same technique'. In patent-lawyer-land the very similar technique you describe will be deemed 'not exactly the same' and thus not valid prior art. Unless you can find a document describing this filesystem which _also_ refers to a similar technique for generating the short names from the long names than you will probably lose the case, because the rules for 'invalidating prior art' are astonishingly strict.

The name-mapping invention described in the patent will remain sufficient to declare it an invention worthy of 20 years exclusive protection, even though we can all see that 90% of it had been done before and the mapping algorithm was hardly a work of overweening genius.

The system is impressively broken, and it's difficult to see how to fix it.

Since Linux's algorithm for generating short names differs,

Posted Jan 15, 2006 15:15 UTC (Sun) by leonbrooks (guest, #1494) [Link]

does that imply that the patent doesn’t touch our implementation?

IIRC (and that’s a long range recall), our implementation doesn’t use the braindead blah~N notation, but guesses more reasonable abbreviations. It would, OTOH, read the blah~N stupidity just as well as its own names (and so would Microsoft's implementation read ours). So is that an implementation of the stupid patent or not?

If it is, then I reckon that we can argue for UMSDOS being prior art for the process. If it isn’t, then the problem goes away.

Goes away for everyone, that is, since other OSes can simply write their own name-guessing algorithms to avoid paying Mr Patent Farmer his “protection” money.