Whatever got conjured up about Invent3k?
September 1, 2010
The 3000 community marks an anniversary this September. A year ago this month, users and vendors were promised that the invent3k public development server, whose 3000 software HP had already shipped to an OpenMPE volunteer, was online and ready to use. At last year's e3000 Community Meet, about 40 people were on hand to hear that on Sept. 23, all you needed was to request an account to use something being called invent3k.openmpe.org.
What's invent3k? A novel experiment from the best of HP's 3000 team, a full load of HP software and disk space to set up development accounts. Considering that HP charged about $100,000 for this collection, making it available for public use was a pact with a community worried about the 3000's future. HP even went on to create an HP 9000 version of the concept. HP shut invent3k down in 2008, but then gave the software to OpenMPE.
Let's get back to last September. Three weeks went by, and then three months, and by the end of six months I'd stopped asking when invent3k would become a community resource once again. I stood in the room at the Hyatt in San Francisco in 2009, reading the slide (shown above) which told us the portal for invent3k was available "through the generous donations of Client Systems and Matt Perdue."
The 3000 hardware? Client Systems sent it. The invent3k software, a full collection of HP subsystem applications like COBOL II? Not online back then at that address, or now. My dismay at the delay turned to despair by this summer, as it became clear invent3k wasn't emerging as a 3000 resource, one managed as promised by Perdue.
It feels good to volunteer when the need is obvious. It can be a tougher thing to volunteer a promise and then keep it -- and perhaps even more of a stretch to describe something as accomplished, but then be unable to meet that pledge.
By a reporter's habits I was recording the events of that largely-upbeat September day. One talk after another fell through my video lens, a collection of proof that the system could still rally a community. But as you can hear for yourself on the 90-second video we've posted up on YouTube, we could only witness one report of invent3k's status. The promise comes from off-camera, but not as far offline as invent3k.openmpe.org has been during the past year.
Around the 23-second mark of our video you'll hear Donna Hofmeister (then a group volunteer), checking up with Matt Perdue offscreen, to verify if his invent3k server was ready to go. His answer, "available now; you gotta have an account" still doesn't fit what what's at invent3k.openmpe.org today: a "not found" as of this morning.
What does this pledge matter to the 3000 customer who's homesteading? Maybe not much, to many of them by now. invent3k was an HP service to the 3000 community, released in a time when having a fast 3000 loaded up with all of HP's development software helped create tools and programs. At the time, no vendor in the enterprise computing industry had hosted a Public Access Development server online. HP pointed with pride at the proof that it cared about the flow of software to the system.
The best use of the invent3k server was as a toolbox for companies that had a less-formal development environment. (Like no crash-n-burn box, or missing some of the HP software.) There's a few members of your community who developed on the HP-hosted invent3k.
Times have changed a lot since then. HP's decisions on its 3000 futures, the closure of the HP labs, departures of developers from the community: It can add up to a summation of "why care?" that invent3k is undelivered a year after we heard it confirmed. In 2010 invent3k is as dark as HP's 3000 lab, and we've never been able to get a straight answer about why.
But don't mistake the change in the community's resources with the legacy of invent3k in your community. Mark Klein started the 3000 open source revolution in the middle 1990s by bootstrapping the GCC development suite, something he built and later polished up using invent3k. Klein reports that his watershed piece of coding -- the keystone to Internet ability and modern networking on the 3000 -- grew up there along with a lot of MPE/iX open source development.
Why bother to say that something is up and running when it's not -- or then later, that it will be real soon? HP's own history with the 3000 is marred with this kind of conjuring, dished out at one key point. Engineers in the 3000 MPE lab kept reporting during 1985 and '86 that MPE XL was running just fine, would be ready for scheduled rollout, even while the OS was keeping debugger screens lit up all over the lab. Eventually the truth about that project came out, like so many other IT project updates that sound good in front of a PowerPoint slide. Slide-ware, or the Foil Effect, we'd call it.If HP can fall prey to this foil effect, why not the independent community of 3000 users? Anyone who takes responsibility for serving others can foil a pledge. But in 1986, HP had millions of dollars to toss into its lab to pull MPE XL out of such a ditch, then get the modern 3000s out the door plenty late. When an independent volunteer can't get a cart out of the ditch, any company watching from the transition fence might accelerate their exit plan. That haste might be the right response, too. Recoiling from the promises of a resource kept in the dark undelivered -- that's a shadow to beat back the bright light of any slide.