TexBlog

Steve Teixeira's Blog -- Thoughts on parallel computing, Microsoft, the industry, and life
Translate This Page
Common Tasks
Search Form
  • Advanced search options...
Tag Cloud
Monthly Archives
Archives

A goodbye to Borland

MSDN Blogs > TexBlog > A goodbye to Borland

A goodbye to Borland

27 Jan 2006 1:43 AM
  • Comments 10

Ken Henderson recently posted this missive, saying goodbye to Borland and providing a few very interesting tidbits about the company's history.  I personally wouldn't proclaim Borland dead at this point.  Yes, it's a bummer for them that Danny Thorpe moved on recently, but it's not entirely unusual for an engineer to seek a new challenge after 15-some-odd years with one company.  Ken brings up many good points in his article, although there are also a few points for which I'd like to offer my own opinion.

I worked at Borland between 1993 and 1998, so I missed the high times and drama of the 80s and early 90s.  However, if there were such a thing as software industry scholars, they would all agree that the acquisition of Ashton-Tate in 1991 was the beginning of the end of Borland as a software industry leader.  It was pretty much downhill from there, as so eloquently expressed in lyric in Dr. Dobb's Journal in September of 1994:

Borland ate
Ashton-Tate
Learned that it was poison, but by then it was too late.

One additional massive mistake that Ken doesn't mention is the price war Borland set off in the office software market by pricing Quattro Pro for Windows at $49 (down from $495!) in 1992.  Competitors were forced to similarly reduce their prices, but the competitors happened to execute better and have more formidable price war-sustaining resources.  It was too late when Borland realized they had brought a knife to a gunfight; they couldn't under-price or out-channel Microsoft's office productivity software.

I also see Anders' choice to leave Borland for Microsoft in 1996 differently than Ken.  Yes, it was obviously not a Good Thing for Borland that Anders chose to move on, but would it really have made a difference if Anders had stayed?  Personally, I believe it would have made little difference in the long run.  The fact of the matter is that the company was already at that time looking for a way out of the developer tools space.  As a result, I don't believe Anders' work was valued by the company the way it should have been.  As such, Anders' departure wasn't so much a "misstep" by the company as an obvious consequence of their strategic direction at the time.

It's also worth discussing the "choice" of Pascal as the language for Delphi.  I've had this asked of me many times over the years.  Why Pascal and not C++?  Look, it's simple: Delphi was Pascal because Delphi was invented by the Pascal team.  Delphi was a brilliant, but very organic, evolution of Turbo Pascal for Windows.  For a variety of technical reasons, implementing a Delphi-like tool for C++ required a great deal more engineering effort, but Borland did make this investment and shipped C++Builder about 2 years after the first release of Delphi.  However, the product never captured the hearts and minds (nor much market share) of C++ developers.

The C++ market in general is another sad episode in Borland's history.  In the early 90's, they were up around 85% share of the PC C/C++ compiler market.  Today I believe they're somewhere in the single digits.  This makes me sad not just because of my Borland history but also because I have a sincere belief that their enthusiastic participation in the C++ compiler market would raise the water level for all of us and ultimately translate into better tools for developers.  Borland seems to be reinvesting in C++ recently, so I'm looking forward to seeing some innovation from them in this area.

The situation at Borland certainly isn't all wine and roses, but then again, this is hardly news; it hasn't been wine and roses since the early 1990s.

Comments
  • wpSlider
    27 Jan 2006 7:39 AM
    I like your post better Steve, let's hope what I type next doesn't get you hot under the collar and result in the deletion of my comment like Ken did.

    Like someone else mentioned in the Borland newsgroups, with so many missteps taken it's actually a wonder that Borland is still with us. But like so many, I am so thankful it has managed to survive because their products just rock!
    C# and VB.NET users have only just recently experienced (PME model, great GUI extensible framework) what Borland gave us back in 1992 - all in native code with no runtime to lug around. With Delphi 8 we were able to compile most well written apps to run under .NET with very little effort. For anyone with a sizeable amount of working code, it's fantastic to be able to target both native and managed code with 1 single tool.

    I don't know why C++ Builder wasn't as that successful. Maybe C++ purists didn't like the Pascal written VCL or the relatively minor language extensions that were added to support PME. Let's hope *your* C++/CLI extensions fare better.

    I hope the old Borland machine still has plenty of life left in it, at least until a Native 64-bit Delphi gets released.
  • Nick Hodges
    27 Jan 2006 11:34 AM
    Steve -

    Good post. Yours is much more reasonable post than Ken's. Ken has added himself to the long and distinguished list of folks predicting Borland and Delphi's imminent demise. I remember Brett Glass leading the charge in the mid 90's.

    (Brett either needs a new picture or a new haircut -- http://www.pcmag.com/author_bio/0,1908,a=221,00.asp)

    Anyhoo -- good analysis.

    Nick
  • Vyacheslav
    29 Jan 2006 1:45 AM
    While I understand why marketing is more important than the product itself, I'm still a programmer, so I feel really sorry for the Borland as its products are really cool.

    For instance, Borland's documentation is still superior to Microsoft's: most of the time I used only help index when I worked with Borland products because every topic had useful tips ad described typical scenarios in the end, and one of the tips usually was what I was looking for.

    Now I have to use Google almost every time I need information about VC++: Google is faster than my LOCAL copy of MSDN.
    If you are working simultaneously with VC# and native VC++ you can forget about filters in MSDN help viewer 8.0.
    I remember how I was looking for the Bitmap::SetPixels function in MSDN index: I found it only when I guessed the function name "SetPixels" myself!

    Another topic is quality. I know one medium-sized project which still uses Borland C++ compiler. They tried to switch to VC++ a couple of years ago and it was runing 2-20 times slower, especially in built-in editor due to different heap managers.
  • 2 Feb 2006 12:50 PM
    spSlider: I don't see anything offensive in your comment at all. I think the Pascal-written VCL was a big part of the BCB dissatisfaction. The language changes were also, but to a lesser extent, I think. The real problem as I saw it was that the product just seemed sort of impotent when compared to Delphi... you could do stuff in BCB, but when you REALLY needed to get down and dirty you had to go to Delphi.

    We're taking great pains not to repeat the language issues as well by engaging the individuals driving the C++ language industry-wide on our C++/CLI bindings. We have already garnered ECMA approval, and we're moving for ISO standardization as well.
  • 2 Feb 2006 12:51 PM
    Nick: Glad you like it. I think Ken has a lot of frustration around "what could have been" like many folks.
  • 2 Feb 2006 12:59 PM
    Vyacheslav, I actually like for Borland to be ahead of us in a few areas because it shows us very clearly where we need to improve. I give Borland props for their docs, and we take documentation VERY seriously, but the size and complexity of our task is several orders of magnatude greater than that of Borland. However, we have some creative ideas coming soon that I think will really help in this regard.

    I invite you to try the VC 2005 compiler. You can even try the free Express Edition available from http://msdn.microsoft.com/vstudio/express/visualC/default.aspx . I would be very surprised if it did not significantly outperform BC++ this time around.
  • Randall Spychalla
    6 Feb 2006 11:38 PM
    I use Visual Studio 2005 professionally and C++ Builder 5.0 personally, and while both have bugs, Visual Studio gives me the most problems (MSDN docs crash, and occasionally the ide itself) that really cut into productivity. Because of that I'm not exactly a VS fan and hope to be able to purchase the newer version of C++ Builder soon.

    I also want to add that I hope Borland considers a stripped down package with just their C++ offering, instead of the whole Delphi/C# package for those of us with less disposable income.
  • Krutten
    10 Apr 2006 9:59 PM
    I have to agree with some of the posts.  Although Borlands shrinking market share and unfocused path (C++ Builder X was a huge step back in Windows development) lead me to believe Borlands days were numbered, I cannot argue with the productivity of there tools.  When I worked in Visual C++ 5 and 6 I had to dip into Win32 to populate listboxes in MFC with 20 lines of code.  In BCB it was one line if I was usng a String List.  Anything I write in Visual C++ I budget twice the time since I will spend a lot of time searching the web and MSDN.  .NET has improved things but Borland still make a compiler and set of libraries that 99% just works right for what you need.   I hope that Microsoft and Borland will continue to work together in the future, but since Microsoft won't let Borland even update the DirectX .lib files, Borland has an uphill battle.
  • 13 Apr 2006 2:41 PM
    Krutten,

    Indeed, BCB was WAY more productive than VC6 for building user interfaces.  Borland and MS have certainly had their differences over the years, but my feeling is that Borland's woes are largely self-inflicted.  Hopefully their selling of the IDE business will mean good things for their tools and the people that work on them.

         -steve
  • Darren John Dwyer
    8 Jun 2006 5:06 AM
    I believe that Borland C++ Builder (or DevCo C++ Builder) offers the best #1 solution for developing applications for Windows Operating Systems...

    If Borland could just look to developing their toolsets to be next-generation applications, then there would be a reason for people to buy their tools.

    For example, not having DirectX support is a pretty big mistake, and even though I personally have created a solution to integrating DirectX and C++ Builder, Borland are still totally lost when it comes to developing next-generation applications...

    Maybe in future, the new company, DevCo, will produce results in the global C++ marketplace.

    I will do my best to help them.

    - darren
Page 1 of 1 (10 items)