Borland has once again "changed course" (I've lost count of how many course changes this makes). Since the suits in the corporate office can't get their asking price for the money-losing DevTools group, they are spinning it off into a separate wholly-owned money-losing subsidiary called CodeGear. Maybe that will make it easier to figure out just how much money they are losing? Meanwhile, what's left of the Borland company is now pursuing 1980's waterfall software metrics. I can't help but wonder how long it will be before they "change course" again.
It appears to me that after over a decade of Borland's attempts to commit corporate suicide, they may have actually succeeded, although the corpse is still thrashing about.
Back in 1983, I saw this wonderful new product called Turbo Pascal for only $20. I bought a copy. It was fabulous -- it had features I had never seen in other software development systems, and I had a useful way to write software on my home system. I became a Borland fan, and I got every upgrade they released to Turbo Pascal.
But 5 or 6 years later, the problems started. Around 1988 or so, they released a Turbo C++. I bought it, even though the price was more than an order of magnitude higher than the original TP. I then discovered that it was a whole lot buggier than the original TP. However, by the third upgrade, they had a useful C++ development system. Then they decided to drop the OWL library, and pretend it never existed. I didn't realize it at the time, but that was a sign of times to come.
When TC++ for OS/2 came out, I bought that, too, and discovered quickly that it was even buggier than the version 1.0 of TC++ for Windows. Shortly after that, Borland silently dropped TC++ for OS/2, and pretended it never existed.
At that point, I pretty much forgot about Borland for a couple of years. But then, I happened to be in a company that was using TP when Delphi 1.0 came out. I saw the demos, and I was truly impressed, and I convinced the management at my client to buy it. Then I discovered that Delphi 1.0 set a new record for bugginess in a commercial software package. I went on to another contract, and forgot about Borland again for a couple of years.
The next time Borland caught my attention was with BC++Builder, which I came in on just after version 3.0. I liked BCB a lot, although it was pretty irritating that Delphi got all the goodies a rev ahead of BCB, and there were some things you couldn't do in BCB at all (like write VCL components). Then Borland "changed course" again. This time, they decided that the thing they needed was a name change, so they changed the name to Inprise. They also decided that they wanted to out-MS Microsoft in the office suite business. About that time, they got a big cash infusion from, of all sources, Microsoft, in settlement of a lawsuit. When I read that MS had gotten a bunch of Borland (er, Inprise) stock in the deal, I began to suspect that Borland (er, Inprise) was on its way to the dustbin of history.
The name change turned out to be about as popular as a turd in a punch bowl, so after a while, they "changed course" again, and renamed the company back to Borland. Meanwhile, they managed to screw up the office suite business so completely that most people today have never even heard of WordPerfect, and I've even forgotten what it was they called their spreadsheet. Having demonstrated a complete inability to make any money in the office-suite business, they "changed course" and divested it.
By the fifth rev of BCB, it got to be a really useful product, although I really liked Delphi a little better. I even joined the local Delphi Developers of Dallas (3D), and did some contracts on the side using it. But there weren't enough people using either BCB or Delphi to justify putting a lot of effort into it, so I wandered back into the world of various Unix systems (and the occasional MSVC++ contract) for a couple of years. There was one new thing that really interested me, and that was Kylix. I played around with it, but couldn't find anybody to pay me to use it, so I pretty much left it behind, too. Just as well, since Borland is currently pretending that it never existed, a corporate trend that I am finding to be increasingly annoying.
After 9/11, it got pretty hard to make a living at software, so I did mostly part-time stuff along with other jobs unrelated to software, so I probably missed several "course changes" from Borland during that period. But I managed to land a full time position with a company that used Delphi 6.0 just recently. I was somewhat surprised to find anybody still using a Borland software development product.
My current employer uses D6, even though there are at least 4 or 5 revs out since D6, and they hired me the day after my interview (my supervisor told me that Delphi programmers are so rare that he always makes an immediate offer -- and he was unable to snag the last few in time before they got an offer to write software in some other language). They have chosen not to upgrade, simply because none of the versions after D6 really added anything they wanted. Plus, since Delphi programmers are so hard to find, and there is basically no support for D6 anymore, they are planning to migrate everything to some other language. Probably C#, since that's the current popular language du jour, and it's easy to find C# programmers.
When I got the job, one of the things I did was look around for Delphi support groups. After a couple of hours of searching, I managed to track down the founder of 3D, who told me that it went inactive after the membership dwindled to about 10 (more than 3 years ago), and that there just wasn't enough interest to justify any effort to re-start it. It looks like my company isn't the only one that refused to upgrade past D6. It also appears that the Delphi user community may have shrunk past a sustainable point.
I still like Delphi, even with its annoying quirks (like the ones that make it tricky to use by large software development teams). I have never seen a software development system that made it so easy to slap a GUI onto a database application, and the compiler is awesomely fast. However, when I looked around for a personal copy of D6, I discovered that it just isn't available anywhere, at any price. But I did find a very interesting open-source project called Lazarus, which is very close to being a D6/Kylix clone. As it turns out, Lazarus is very nearly as useful as D6, and is rapidly improving, so I'm planning to try it out on a little side project for an old client. And once my current employer manages to migrate away from D6, I don't expect to encounter Borland (or whatever they decide to call what remains of it) again in my career.