FAT32 should be a QA-ed, RAND-z standard

By Rick Jelliffe
March 11, 2009 | Comments: 8

Hobby horse time again!

One of the basic reasons I supported the development of an ISO standard for OOXML was my belief All market dominating interface technologies should be QA-ed, RAND-z standards. Once you have achieved a position of market dominance for a technology, the market for selecting between technologies dries up and needs to be replaced by a lively market for symbiotics and substitutes that use that technology. Standards encourage this. I think this is a position that many competition watchdogs are sympathetic to.

The FAT32 file system is surely another example of this kind of market dominating technology.

Like OOXML, it needs to be brought into a position of being a market-creating technology rather than a market-stifling one. Submission to a standards consortium with clear open IP access rules (e.g. Ecma in the first instance) would seem entirely appropriate, and I am sure would win a lot of continued goodwill.


You might also be interested in:

8 Comments

Mr. jelliffe

sorry , but i have to tell you that Santa Claus is not real.

Cheers to Microsoft

Carlos: It is real with enough pressure and government intervention. Hope springs eternal.

People said that MS would always keep its file formats close to its chest, but has been moved. This is just another of these issues.

I don't see that there is any difference between file formats, APIs and file allocation table formats: they are all interface technologies.

They are the kind of thing that Microsoft should be applying their Interoperability Principles to. Now perhaps the current legal case is actually about different issues, but it is being reported as being about FAT32 usage: but my argument is not because of anything to to with the merits of that case, nor even about consistency with their Interoperability Principles.

It is on the use of open standards as a tool to stimulate new markets/bazaars on top of markets/bazaars stifled by proprietary dominant technologies that will not magically go away, and in the particular broad case of "interface" technologies.

Fostering workable markets/bazaars is hardly airy-fairy: it is the central economic issue.

FAT32 is a 15 years old technology, build on top of a 30 years old technology, and is not ideal at all as a file format for SSDs. I don't see how it can be a market-creating technology, and I really don't see the interest of making it an ECMA standard, even less an ISO standard.

If you want to support a market-creating technology, you should rather recommend to make e.g. LogFS or UBIFS an open standard (they are already open source technologies).

This being said, it would indeed be good to push Microsoft to provide FAT32 on RAND-z basis, as they have a de-facto monopoly on flash drive file systems, so that they are not tempted to abuse it.

Luc: If there is interest in LogFs or UBIFS, then certainly someone could champion them to ISO or some other standards body: Ecma is particularly good at processing pre-existing contributed technologies, for example.

I think the difference between us is that I see standards as a practical method of getting standards QAed and RAND-z, while I guess Luc does not: the fact that something is old and not theoretically ideal is not a bar for standards: the essense of a standard is agreement and disclosure.

Voluntary disclosure is OK, but there needs to be some invigilation, which standards bodies (with adequately broad and untied membership) can provide.

Don't you see a connection between your comments on FAT32 not being a market-creating technology and your comment on it being monopolized?

From the FOSS POV, all the dominant interface technologies (file formats, file systems, protocols, APIs) need to be open, and standards are the best way for this: the argument that this will entrench them further is no argument when they are already entrenched!

The FAT patents from MS are already available on RAND licensing.

hAl: The FAT32 spec is at
http://www.microsoft.com/whdc/system/platform/firmware/fatgen.mspx

I am including VFAT with FAT32, by the way.

The original FAT patents were rejected in court:
http://www.internetnews.com/bus-news/article.php/3415871

According to Wikipedia, the patents on VFAT (long names) and FAT32 are current.

RAND is not good enough for these technologies. It has to be RAND-z *and* QAed (*and* with any other IPR cleared up.) OSP (is that what it is under?) is good, but not enough.

Rick,
> If there is interest in LogFs or UBIFS, then
> certainly someone could champion them to ISO
> or some other standards body
Indeed a standard based on LogFS and/or UBIFS would be a good thing. My understanding is that Ecma is rather targeted to industry standards, and recently lost much credit in the eye of open source advocates, but it is surely possible to find an adequate venue for such a standard.

> I see standards as a practical method of getting
> standards QAed and RAND-z
Standards are surely a good way to have technology QAed but FAT32, due to its age, is implemented in virtually all devices exchanging files, and it is quite late to make changes to it. And as far as I know, neither Ecma nor ISO asks for technologies submitted as standard to be RAND-z. So indeed, we have a different view about this. I think that energy would be better used to create a new standard well adapted to the current needs, in the same way as the PNG standard was created when patent issues appeared with the GIF format, both to avoid the patent issues and to overcome the GIF limitations.

> all the dominant interface technologies (...)
> need to be open
I fully agree with you: we need a good, open, standard technology for SSD/flash file systems.

> the argument that this will entrench them further
> is no argument when they are already entrenched!
I agree again. This is not my argument against spending time to standardise FAT32. My argument is a technical one: FAT32 is a very bad file system for SSDs (especially big ones) due to the way the flash technology operates. See for example http://lazybastard.org/~joern/logfs1.pdf.

Luc: Ecma is perhaps a good body because it allows free ($$$) participation by a broad range of people (without requiring special permission) and it is quite geared to coping with hardware.

On the other hand, final votes will undoubtedly be dominated by corporate membership. So in order for OASIS to work, you would have to get hardware makers who are members of Ecma onside. But is a standard for flash file systems practical without hardware-maker support in the first place? I suspect not, and it is this kind of coallition-building that is important for standards.

I don't think OASIS has any standards with a particular hardware angle, or hardware manufacturers, and certainly it is out-of-scope for W3C. There are probably other bodies possible too.

On the issue of FAT32/VFAT, having a standard for them does not imply that there could not be a favoured standard for a better technology. The QA on standard for an existing technology is more a QA on the documentation rather than alterations to the technology after all. (E.g. an ISO standard for the Linux ABI could not change the Linux ABI willy nilly without it becoming the standard for some imaginary technology.)

News Topics

Recommended for You

Got a Question?