Client Server NEWS 361 - August 7-11, 2000, InterBase Flap Could Fork Code

By Maureen O'Gara

As featured in Client Server NEWS 2000
The Independent Observer of Microsoft Windows NT and Other Phenomena
Sign up for a free three week trial subscription:
(c) Copyright 2000 G2 Computer Intelligence Inc. May not be forwarded by e-mail, reproduced, stored in a retrieval system, posted on an Internet/Intranet web site, or transmitted in any form or by any means, electronic, mechanical, photocopying, printed, recording, or otherwise without prior written permission of G2 Computer Intelligence Inc.
Client Server NEWS 2000 is published weekly by G2 Computer Intelligence Inc.
323 Glen Cove Avenue; Sea Cliff, NY 11579; USA
Tel.: 516 759-7025
Fax: 516 759-7028.
Send press releases to
Comments? Subscription info?: e-mail: or call 888-809-6397

Inprise or Borland or whatever it calls itself these days was supposed to open source InterBase and then turn it over to a company that InterBase's original developers were putting together.

It got as far as open sourcing the code for the end-of-life database that in its heyday was an Oracle wannabe. Oh, okay, Sybase wannabe.

The company also open sourced the Linux, Windows and Solaris binaries for InterBase 6.0, the latest rev, same rendition as the source code it released.

Everything is out under a variant of the Mozilla license, which means developers don't have to open source their modifications or InterBase6-based applications. Mozilla makes InterBase different from the MySQL database, for instance, which is going GPL, because GPL would require MySQL apps to be open sourced. (See

That was the week before last. At the end of last week Inprise publicly pulled the plug on the negotiations it had been having with its InterBase general manager Ann Harrison and the projected InterBase Systems (ISC) spin-off since at least St Valentine's Day. Harrison, who walked out of Inprise when the negotiations hit the wall, was one of the original gang of four that created InterBase. She is married to Jim Starkey, the original founder of InterBase, who sold the company after seven years to Ashton Tate. Starkey, who by his own admission gave up recommending InterBase a year and a half ago because of the "gross incompetence of Borland management," was to serve as ISC's technology software architecture advisor. Inprise, by the way, formally announced ISC's formation in mid-February.

Late last Friday, Inprise's interim president and CEO Dale Fuller said in a canned statement that, "After careful consideration, we determined that it was not in the best interest of Inprise/Borland's stockholders for us to sell the InterBase product line to a start-up entity which initially would be dependent on Inprise/Borland for funding." Inprise said it was going to support the open source effort itself instead and was retaining a core transition team.

Inprise had originally said it would underwrite ISC from its venture capital fund.

Inprise senior VP of business development Ted Shelton said in a series of increasing revealing but not completely satisfying e-mails to developers, many of whom feel Inprise acted in bad faith, that, "we were not able to come to an investment agreement with Ann Harrison and Paul Beach," ISC's VP of sales and marketing and that "we made offers to Ann that she did not accept, and that Ann made offers to Inprise that we did not accept."

In a more expansive message on Sunday, Shelton said, "Ann and ISC proposed to have an independent company from Inprise and asked for the technology, trademarks, and for a venture capital investment. There were, in addition, a number of legal requirements that they placed around the transfer of these funds and the intellectual property. A significant effort was made to reach a middle ground in our negotiations with Ann. However, we were not able to bridge the differences in opinion. As we have an obligation to our shareholders and employees in addition to our obligation to our customers, we could not simply agree to ANY conditions that a third party was going to put on a transaction of this kind." Inprise had originally conceived ISC as a company to service, support and host InterBase 6. At least that's the way it described it in its St Valentine's Day press release.

Shelton also said that Inprise is "working on a plan around supporting InterBase as an open source product" and that Inprise would "welcome the development of new companies seeking to support InterBase."

Fuller has made a lot of people unhappy with his decision and not the least of them are Harrison and Starkey, who might very well go ahead and form ISC anyway and fork InterBase.

Starkey and Harrison, who on Friday, after the axe fell, continued to describe herself as president of "NewCo," the re-dubbed ISC, disappeared at the weekend - at least they're incommunicado and not answering e-mail - and are evidentially considering their options. The InterBase community appears ready to follow them down the NewCo path, judging from the postings to the public mailing lists. NewCo may be restyled Firebird and InterBase or rather the InterBase clone given another name and tag line.

Starkey in mid-July blamed Inprise for a lack of resolve in open sourcing InterBase and getting ISC off the ground. "Deal after deal has been lost," he said of the delays. "Potential investors have been turned away. Important customers have drifted away as Borland's lawyers missed deadline after deadline."

Even without Harrison and Starkey, who have been with the product for 20 years, the community could fork the database. Inprise didn't release the InterBase 6 documentation - written partly by Inprise and partly by ISC - cross-platform tests, build scripts or bug lists. (It's rumored Inprise wanted ISC to pay $10 million for the stuff.) The community could surely reproduce them and fork the source. Some enterprising folks have already figured out how to re-build it, at least on Linux. There are notions about hosting any schismatic effort on VA Linux Systems' SourceForge site. Anyway, there are moves to set up a replacement web site, get the source tree organized, put documentation together and synchronize the code.

Inprise has also alienated its VAR community, which first started to drift when Inprise decided to kill InterBase off in December - an idea it recanted when word leaked and a tempest ensued. Then there are the people who have been working for free on little items like drivers under the impression that they'd get a piece of ISC.

Angry developers says Inprise released "unbuildable source code, broken binaries and [is] withholding the pieces that members of the Open Source community need to actually give this product a life."

Meanwhile, Inprise went ahead Monday and announced an InterBase 6 deal with Cobalt Networks that was shaky a couple of weeks before. Cobalt is going to bundle the relational database with its RaQ 4r server appliance. Cobalt imagines InterBase applications will be "appliantize'd" and pre-configured with its Linux-based platform. InterBase 6 is largely cast as an embedded RDBMS since it failed as an Oracle contender.

Paul Beach takes credit for putting the Cobalt deal together, an 18- month effort at relationship-building. Inprise was supposed to release InterBase 6 binaries to Cobalt by July 14 but Fuller, who had reportedly promised to release the code, for some reason turned the issue over to his lawyers at the last minute. They refused permission for the code to be released. At least that was the story given out.

At the time, Starkey sent around an e-mail saying, "Unless Borland agrees to release version 6 binaries to Cobalt today, July 14, the Cobalt deal is probably dead. If the Cobalt deal dies, the Borland/InterBase deal may die with it. In any case, if the Borland/InterBase deal does not close by July 24, the deal will be dead. The future of the product will be Dale Fuller, not Ann, Paul and Matt. I will not be involved in any capacity." Well, Beach apparently salvaged the Cobalt deal and Starkey was right about the InterBase arrangement.

InterBase 6 introduces new features including new data types (long integer, data and time), extended SQL92 compliance, an open interface for defining new national character sets plus performance and security enhancements.