Archives of the TeradataForum

Message Posted: Tue, 25 May 2004 @ 13:50:03 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

<-- Anonymously Posted: Tuesday, May 25, 2004 03:10 -->

I missed the begining of this thread (my laptop is dead -again-), but I can add two things.

One, there IS a Linux client version available for Teradata. The standard release of TUF 7 includes the CLI libraries for Linux, burried somewhere on the CD. There is a "complete" client port available from the customer porting group: you'd need to check with sales for that availability and pricing. I've been running personal CLI aps at home against my TNT servers (r4 & r5) without incident...although I do miss Queryman...

Second there is a persistent rumor that Linux is somewhere in the roadmap for the database platform. I'm a field guy, so I usually learn more from this forum than I do internally, but I've heard that this may be underway, if only as a "hobby shop" task by the developers.

Also, with regard to GPL/OpenBSD licenses: first, the BSD license differes from the Gnu in that you -CAN- incorporate open code into proprietary without requiring full disclosure of the proprietary contents. This isn't as onerous as it seems: it effectively "dead ends" a proprietary extension, while the open source equivalent will continue to grow with contributions. But it does give companies the ability to provide a way to protect their IP without completely ignoring the open source community.

If you don't incorporate GPL code into an application that uses an API from a GPL application, you're not violating the GPL license. This is sort of the Microsoft IP licensing issue in reverse: you have to promise that you didn't steal anything that was in the public domain. If you built binary code that linked to the GPL kernel without requiring a source inclusion, you could effectively extend the kernel without violating the letter (if not the spirit) of GPL licensing. Again, you lose the advantage of the open development community, and if someone develops the same thing using the open source, you are effectively bypassed, but it becomes a business decision to pursue this. Teradata is a good case for doing one of these approaches: the target market for Teradata is not a general purpose tool replacement (like OpenOffice), but a fairly small market with specific needs.

Besides, as an earlier post mentions, the stand-alone demo can be provided without any license issues. And, for MPP systems, you still need to purchase BYNET hardware to make the cluster work...effectively making even an open source implementation "proprietary".

Now, if someone were to purchase a complete 4 node system on the open market, completely strip MP-RAS out of it, and develop a reverese- engineered open source BYNET interface strictly from open specifications...Hmmm, isn't this similar to the controvery about Linus Torvalds and the Ken Brown "report"?

NOTE: ALL COMMENTS IN THIS POST REFLECT PERSONAL OPINION AND ARE NOT THE EITHER THE OFFICIAL POSITION OR OPINION OF TERADATA-NCR, ITS MANAGEMENT OR STAFF.

	<Prev	Next>		<<First	<Prev	Next>	Last>>