Hans de Goede
j.w.r.degoede at hhs.nl
Thu May 31 12:48:11 UTC 2007
James K. Lowden wrote:
> A decision was made back in 2005 not to include FreeTDS in Fedora Extras.
> The decision's reasoning rested on faulty logic. I am not an RH user or
> developer; I'm the FreeTDS maintainer. I would like to clear up the
> misunderstanding today and request FreeTDS be included in some way in the
> Fedora distribution.
> The above cited message suggests FreeTDS is under a legal cloud. The
> facts are:
> 1. The TDS protocol has been public information for many years.
> 2. Microsoft's patent 5,974,416, mentioned above, does not apply.
> 3. Microsoft and Sybase are well aware of FreeTDS and have never
> suggested any kind of infringement exists.
> 4. FreeTDS serves the business interests of both vendors.
> Sybase's document
> ) is freely downloadable. It is marked "confidential", true, but it has
> continuously been published on their website for several years, so any
> confidentiality is surely now minimal. The mere fact that something is so
> marked, in any case, has no legal import. Nothing else in the document
> suggests its use is in any way restricted.
> The assertion that FreeTDS violates the LGPL license that governs its use
> is incorrect. The TDS protocol is not covered by any patent.
> Microsoft's patent 5,974,416 does not mention and does not describe TDS.
> It issued in 1999, 15 years after Sybase invented TDS, perhaps five years
> after Microsoft licensed it from Sybase, and two years after FreeTDS was
> first released. Yes, the patent has words like "tabular" and "data" and
> "stream" in it. No, TDS does not have any of an "Advanced Data TableGram
> (ADTG) format", a "HandlerOptions section", or a "Descriptors section".
> There is no "Descriptors section [that] comprises a 16-byte unique
> identifier." In short, the patent describes something similar to TDS that
> the inexpert reader might be led to believe is the same thing, but which
> in fact specifically avoids language that describes TDS. This should come
> as no surprise: Sybase surely would object to having its protocol patented
> by Microsoft. One might even suggest the USPTO would object, too.
> The assertion in a reply
> "There are 4 versions of TDS: 4.2, 5.0, 7.0, and 8.0. MS bought TDS 4.2
> from Sybase. Only Sybase uses 5.0, and only MS uses 7.0 and 8.0."
> is true but unimportant. These flavors of the protocol are much more
> similar than different. Only a small part of FreeTDS distinguishes
> between them. The features Microsoft added are limited to a few new
> datatypes, the use of UCS2 encoding for metadata, and some minor
> annoyances. No one would plausibly claim that TDS 7.0 represents an
> invention of any kind, much less a protocol distinct from its predecessors
> in any but the most technical way.
> The FreeTDS project has been contacted by parties at Microsoft from time
> to time. Discussion at one point led to how Microsoft might support the
> project in some way. If patent violations were at issue -- which clearly
> they can't be -- surely those who opened the discussion with us would have
> been warned off. In the end, though, nothing came of it; they would
> have required a nondisclosure agreement no one on the project was
> interested in signing. (IMO the LGPL was the real stumbling block.)
> Finally, a Google search for "freetds site:microsoft.com" quickly shows
> that Microsoft actively supports FreeTDS to a surprising degree. Microsoft
> makes no Unix drivers of any kind. I do not think I have to explain how
> Microsoft benefits from the existence of a free driver connecting
> Apache/PHP sites (to cite just one example) to their database product.
> The facts are all on one side: the documentation, the patent language, the
> age and history of TDS, and the vendors' behavior and interests all tell
> us that TDS is completely, absolutely unencumbered. It cannot be patented
> or copyrighted, and it is obviously not a trade secret. It doesn't even
> stand vaguely *accused* of violating patents. FreeTDS is, simply, free
> software. I wish the Fedora project would accept and treat it
> Thank you for your time.
Thanks for your long mail!
Notice that freetds is currently in livna already, but having it in Fedora
proper would be advantagous for packages like libgda (which could then support)
tds) and php-freetds
This gets a +1 from me.
Spot, does this need to go through RH-legal, or can you ok it?
More information about the fedora-devel-list