Weird developments
Open source and open sores.
by Esther Schindler
The last few months have taken OS/2 users on a roller-coaster ride. We have some upbeat news, including the announced availability of StarOffice and Watcom compiler source files. Yet, we also have some news that bodes ill for the OS/2 development community, and a continuing attitude from IBM that I can only label "active neglect."
Let's talk about the good news, first.
Watcom open source
Sybase, Inc. announced that it will release the source code for the Sybase Watcom C/C++ and Watcom Fortran compilers under an Open Source license agreement. As you might recall, Watcom was purchased by Powersoft, which was thereafter acquired by Sybase. The Watcom family of compilers were shelved into "no further development" status not long after Sybase took over.
According to the press release, this move will "allow existing customers to continue to evolve the Watcom compiler products for their own use."
SciTech Software, Inc. will be the official maintainer of the Watcom compiler source code, which is planned to be released under the "Open Watcom" moniker. According to the release, SciTech was selected due to its heavy use of Watcom products, its extensive cross platform experience and its record of Open Source involvement; SciTech released its commercial SciTech MGL graphics library to the Open Source community in 1997. SciTech may also be familiar to you as the author of the SciTech display drivers; they're quite familiar with OS/2 issues and the OS/2 community.
The Open Watcom site will host the project source code, bug tracking database, developer discussion groups, and official binaries for free download. SciTech will also be heavily involved in continued development of Open Watcom products. The first project is to create and distribute a binary patch upgrade release (11.0c) with bug fixes developed by Sybase since the last commercial update. Afterwards, the source for the Open Watcom compiler products will be released on the Open Watcom Web site, so that any developer will be able to use and modify the code.
For us, this means only good things. The availability of a "new" C/C++ compiler (and Fortran compiler, to a lesser degree) will enable OS/2 programmers to create and update applications. Watcom's development tools have always had an outstanding reputation; Watcom's C/C++ were used as the "fine finishing" tools before an application went out the door. Plus, it's especially heartening to learn that cross platform tools will be in the hands of people who care about them, under the supervision of a company that has a strong and recent history of supporting OS/2.
StarOffice open source
Sun Microsystems announced, in July, that it will release StarOffice 6 under the GNU General Public License on October 13. This is only a spot of good news, however, because Sun is oddly uncommunicative about what they'll do with earlier versions--particularly the 5.1a version, which was the last one released for OS/2.
Sun created and is funding an independent OpenOffice.org Foundation modeled on the Apache Foundation. Sun will hold a minority position, while Collab.Net coordinates the StarOffice community. Sun, which retains the StarOffice brand, will ship an implementation based on the OpenOffice.org source tree, and StarPortal, a hosted version of StarOffice.
However, the last version for OS/2 wasn't 6.0. What does that mean for us?
To clarify the situation for OS/2 users, I corresponded with Amanda Carter, a Sun Customer Service Specialist. Unfortunately, Carter's messages shed little light on the subject. I wrote to Carter saying that I was given to understand that Sun's release of Star Office won't apply to the OS/2 version. If that were correct, could she please explain why?
She responded with a cut-and-paste explanation: "As part of the long-term planning for StarOffice, Sun has assessed platform demand, momentum in the marketplace, the cost of porting to those platforms, as well as the business case for adding or dropping platforms. Based on this analysis, Sun will no longer release future versions of the StarOffice Suite on OS/2. Version 5.1a will be the last version of StarOffice on OS/2. Sun will provide a range of support and transition assistance to OS/2 customers. StarOffice software will continue to be available on the Solaris operating environment, Linux and Windows 2000/98/95/NT."
That's a complete answer, but unfortunately it didn't match the question I asked. I wrote back, pointing out that her text was old news. However, if Sun is releasing the source code under the GPL, then they no longer have to worry about "the business case" for the platform. If the source code is available, the OS/2 community can choose to provide that support itself. Yes? Or was there something here that I'm not completely understanding?
Carter responded with another batch of boilerplate text: "StarOffice Source code is scheduled to be released on October 13th, 2000. It will be available at www.openoffice.org."
Oh. Well, I guess that means we get to work with the 6.0 source code, just not the OS/2 version thereof. At this point, faithful reader, your guess is as good as mine. (I should also have contacted Sun's PR folks, but the last few weeks have been especially busy for me. I apologize.)
Whatever the details, however, the StarOffice open source release could be a good thing for OS/2 users. If someone picks up the development ball for StarOffice 6.0, it's possible for us OS/2 users to get that upgrade after all. In any case, this isn't bad news.
The situation gets somewhat more dim from here on out.
VisualAge goes blind
IBM announced that "Program Services end date" for IBM VisualAge C++ compilers (for OS/2 and Windows NT) is January 31, 2001. There are currently no plans for any new versions on those platforms.
This isn't an OS/2 issue, clearly; the Windows version of VisualAge C++ is affected, too. Of course, this won't affect Windows developers overmuch, since they still have several C++ compilers to choose from. Until the announcement about the Watcom C++ compiler release, many OS/2 developers were getting rather disgusted. Although, considering the low opinion many programmers had of the last VisualAge C++ for OS/2, from a technical perspective this may not be much of a loss. I'm reminded of a story that Woody Allen told about two women visiting a hotel in the Catskills. One says, "The food here, it's poison, you can't eat it." Her friend agrees, "Yes! And such small portions!"
But don't imagine that the end-of-life equally distributed to Windows and OS/2 for the C++ compiler is evidence of IBM evenhandedness to OS/2 users. It's depressingly obvious, to this most casual observer, how much IBM wants to sweep OS/2 under the rug. Here's one example: at the recent IBM Partnerworld Developer's Conference in Las Vegas, I noticed a pedestal devoted to WorkSpace On Demand. I chatted with the IBMer staffing the booth, for quite a while, and learned that he'd spent years working on OS/2. When we talked about the small collection of workstations he had, demonstrating WSOD, he told me that he had another Thinkpad, set up to boot OS/2 off the server. "They told me to take it down," the IBMer said.
More annoying, in many ways, is that IBM also announced that it will no longer update the OS/2 version of VisualAge for Java. I find the message contradictory, at best.
On one hand, IBM NCSD—and John Soyring personally—have said that OS/2 users should choose platform neutral applications and create those applications using non-platform-specific tools. OS/2 customers should move in the direction of "network computing," choosing tools that will migrate with them to whatever platform they choose. Java is a prime example of that "platform neutral" paradigm.
But then IBM says that developers won't have tools with which to create those platform-neutral applications. Doesn't that undermine the strategy message? If new versions of Java development tools aren't available for those OS/2 developers, exactly how are these people to make their migration? That's like burning the bridge before you cross it.
In this case, I did correspond with an IBM press relations person. Rather than interpret Tim Keeley's response, I'll reprint it; make your own evaluation.
"Stopping IBM development of AD tools for hosting on OS/2 and withdrawing support for AD [application development] tools on the OS/2 platform is entirely consistent with our OS/2 strategy for e-business. John Soyring, Vice President, IBM e-business operating systems solutions, has been advising OS/2 software developers since 1995 to use Windows NT, AIX or (recently) Linux, as their development platforms.
"The reason is that the largest set of commercial AD tools is available on these platforms, and vendors who create new or enhanced tools are developing for these platforms. Very few AD tool developers, if any, develop their tools to be hosted on OS/2 as the number of software developers using OS/2 as their development platform is too small to justify an investment.
"At the same time, using tools on other platforms to create applications which run on multiple platforms (including OS/2 Warp 4 or OS/2 Warp Server for e-business) is entirely consistent with IBM's strategy for OS/2... and consistent with what John told you when you interviewed him on April 24.
"This approach fully supports IBM's Application Framework for e-business. Net: Our messages are not contradictory."
In point of fact, John Soyring did not say that OS/2 users should develop on other platforms. He has never said that to me, once, in the several conversations we've had since 1995 (or since 1992, when we first met). He has said repeatedly that OS/2 developers should use cross platform tools, particularly Java—but Keeley's statement does not match my experiences in the least.
You know, I'd have accepted their news graciously, if they claimed sales too low to justify development. I wouldn't like it, but I'd accept it. For IBM to claim that this is consistent with their policy really irks me.
Have you seen differently? I'd like to hear about it.

