February 1, 2005
Borland's quality problems and Delphi 2005Posted 3949 days ago on February 1, 2005
In the November/December issue of Application Development Advisor I wrote a column on Borland's Delphi 2005. Basing my words on a late beta, I was generally positive, and I still think it has a lot to offer. You can target .NET or Win32 from the same IDE and with the same RAD language; there is integrated refactoring; and there is strong modeling support. Borland has also done some great work on the Delphi .NET compiler. Informal tests suggest that its code generally performs better than C#. The GUI framework in VCL.NET is also a lot brisker than Windows Forms, though it presents tricky compatibility problems and it is probably better to stick with Windows Forms, which Delphi also supports.
What I didn't realise when writing the review is that Borland would release the product before it was ready. I have personally come across several people who tried the shrink-wrap and found such problems installing, or such unreliability once installed, that they have put it back on the shelf. Borland has released one quick update, but there are still major problems. In particular, the integrated modeling support in the high-end edition is a disaster. Memory usage shoots up, performance declines drastically, and crashes are frequent. Developers have even figured out how to hack the install in order to remove the modeling support and make the product usable. Some Borland guys, already mistrustful of .NET, now think that the entire .NET platform is buggy and unreliable. Others look at Visual Studio .NET and realise that it is possible to build a reliable IDE. I don't mean to say that VS.NET has no issues whatsoever, but compared to Delphi 2005 it is highly dependable.
At heart Delphi remains a fantastic product, particularly for Win32, and I still hope it can make its way in the .NET world. What puzzles me is why Borland chose to release Delphi 2005 in a near-unusable state. It is hugely damaging to long-term sales. It is not the first time either: look at Together for Visual Studio .NET for another example (I realise there is a common thread here).
So what should a software vendor do if the pressure to ship (from financial and marketing departments) builds to an irresistible force, never mind that the product isn't ready? I'm not saying this is what happened at Borland, but what if? The answer is that it's better to cut features than to ship something broken. Promise the missing bits in a future free update if necessary, but get the quality right from the beginning.
My supplementary question is whether I was wrong to write positively about the beta. I generally don't comment on bugs in beta product because it does not see fair or relevant. Is it better to wait for the release and live with the resulting delay? Feedback is welcome.
Comments are closed
Recent postsUsers plead with Borland to give up .NET
IE7 to be released 18th October,...
If Microsoft doesn't use UAC, why...
Google's unsettling lack of direction
Vista security: now prove it