Skip to main content
Thoughts from David Cornelius


Delphi is one of the greatest development environments every produced for Windows. It has an easy to learn, yet strict language that leads to less confusion than C++ and better coding practices than Visual BASIC. Unfortunately, it has been marketed by a company that has made so many changes in direction and name that people have laughed it off. One more change has happened recently, here is the story, which actually starts over 25 years ago, before Windows.

In November, 1983, Borland released Turbo Pascal 1. It was an inexpensive DOS-based Pascal compiler written by Anders Hejlsberg in Denmark and only produced .COM files. But it did them well and did them fast. Hobbyists and students snapped up copies, and the phenomenon was born.

I started using Turbo Pascal 4 professionally in 1988 and in a few years, Borland renamed it to Borland Pascal, upped the price, and entered the professional developer market, competing directly with Microsoft. With Borland Pascal 8, you could use OWL (Object Windows Library) to create 16-bit Windows applications. But it was a huge library and not many were brave enough to tackle it.

Fortunately, there was a new and much better solution coming. In 1995, Borland released Delphi 1, the first development tool to use the concept of visual components placed on a form to encapsulate library code--which is now prevalent in nearly all IDEs. The next year, they released Delphi 2 to address the 32-bit capability of Windows 95; this version also added visual form inheritance.

Delphi 3 was a solid product and added some nice IDE enhancements, but more importantly added support for ActiveX and introduced WebBroker.

Delphi 4 came during a time when Borland was entering some troubling years. The company was renamed to Inprise Corporation, supposedly to align it with more enterprise-level customers and break into more application-life cycle products. The developer community was not happy--especially since Delphi 4 was so buggy. However, the IDE continued to break new ground and add new features that have been adopted by other companies.

In 1999, Delphi 5 was released and was a very strong product (still in use today by some hold-outs). Again more features and IDE enhancements were made, including support for ADO. By this time, everyone realized that renaming the company was a mistake, so the product was branded as "Borland Delphi 5--by Inprise Corporation."

With the new millenium, came another version, Delphi 6. Many changes to the compiler broke backwards compatibility, but allowed for cross-platform development with Kylix, the company's Linux experiment that died after only 3 versions. But more significantly, support for web services was added and many other internet, COM, ADO, database, and IDE features and enhancements made into the product.

Many say that Delphi 7 was the last great Delphi. It used the same familiar IDE that existed in Delphi 1, used the fast WinHelp online help system, and was faster and more solid than Delphi 6. The company had changed its name back to Borland by now and developers were very happy (and productive) with Delphi.

But Microsoft .NET (architected largely by Anders Hejlsberg who left Borland in 1996) was here and Borland had to address the new technology. So they added a .NET preview compiler as part of the "studio" package for Delphi 7 on the 8th anniversary of Delphi.

Delphi 8 was the "real" version of the .NET compiler for Delphi. In fact, if you bought the "studio" package, the Win32 compiler part of the deal was actually Delphi 7. However, Delphi 8 had so many problems as to be practically unusable.

Delphi 2005 finally married both Win32 and .NET development in a single IDE, and even though it was still terribly buggy, at least it worked. Borland made huge efforts to keep backwards compatibility to help bring Win32 devlopers over to .NET by providing VCL for .NET. This actually worked amazingly well--for the most part.

One of the biggest complaints of Delphi 2005 was the radical departure from the fast and familiar WinHelp to the new Microsft Compiled HTML Help format. Even though this new format was more flexible, it loaded far slower. Worse yet, very little of the old help text had been ported over to the new format and resources at Borland were stretched to the max for supporting several different products, technologies, and newly acquired companies.

With a buggy compiler, a slow and inadequate help system, and Borland's increasing interest in ALM and decreasing interest in developer tools, many customers around the world began looking at other solutions, or simply did not upgrade from Delphi 7. Confidence in the company was shaken and many questions were raised about the future of Delphi.

Borland continued to promise support of Delphi and delivered that to some extent the next year with a much improved Delphi 2006. The enterprise versions of the "studio" package included requirements, modeling, version control, testing frameworks for both Win32 and .NET, multiple database support, installation, and of course all the languages: Delphi Win32, Delphi for .NET, C++Builder, and C#Builder! It was quite an impressive package, but still developers were uneasy and asking hard questions.

The grumbling about the departure from Delphi 7's efficent editing never seemed to stop. But other concerns were escalating. With Borland acquiring more ALM companies and putting out buggy products with incomplete documentation, developers couldn't help but wonder if the speculation over the years about Borland's eventual demise would finally come true. Or if it stayed afloat, would there be enough support for them? Many jumped ship, but others stayed on to see what would happen.

What did happen was quite surprising to many. In February, 2006, Borland announced that it was going to sell off the Developer Tools Group, the team that worked on all the development and database tools. Lots of speculation ensued--and the future was still uncertain. (The original announcement has been taken down, but an InfoWorld article still exists that talks about it.)

The DTG started "evangelizing" Delphi by offering tutorials, soliciting customer feedback, and even reviving the old DOS-day "Turbo" name by offering Turbo Delphi, Turbo C++ and other "Turbo" products for free. The products offered had some limitations of course, and were basically the single-product versions of Borland Developer Studio 2006, but probably did some good in bringing back fond memories, or at least reminding people that Delphi had some serious history. (I'm not sure if the key reason for the Turbo products--to raise awareness for Delphi and bring in new customers--was terribly successful.)

A new level of excitement arose (albeit mixed with trepidation) and in November, 2006, Borland officially formed CodeGear as a wholly owned subsidiary from the Developer Tools Group.

Many saw this as a very positive sign that not only was the group of people working on the products going to stay, but they would manage their own company and focus solely on developer tools. To answer this, CodeGear released Delphi 2007 only four months later. While mostly a bug fix for Delphi 2006, it introduced a few nice IDE features and Blackfish SQL (a revamped JDataStore). But more importantly, it was a sign that the new company was active and committed to producing a quality product which addressed the customers' concerns.

A little over a year later, Embarcedero purchased the CodeGear division from Borland. Delphi had finally left the company that took it from learning tool, to serious programming language, to enterprise-wide development platform. But even with a full suite of languages and platform support and now with the backing of a company who really wanted the products, the story had another twist coming.

In 2008, Delphi 2009 for Win32 was released with unicode support--the most significant upgrade ever for the Delphi line. Strangely missing was the .NET component.

Meanwhile, another company, RemObjects (which had produced several Delphi libraries) had written the only other Pascal compiler on the planet (well, since Microsoft Pascal for DOS in the early 1990s), but instead of providing a rich development environment like Delphi, built it as a plug-in to Microsoft's Visual Studio IDE. This product was named Chrome. There was no upgrade path from Delphi Win32 or Delphi for .NET, but that was by design--they didn't want to bring along lots of historical baggage. Their release for Visual Studio 2005 had been slowly gaining ground.

Then in October, 2008, RemObjects and CodeGear jointly announced Delphi Prisim which was the Delphi IDE coupled with RemObjects Pascal compiler for .NET, or version 3 of Chrome that works either in Delphi or as a plug-in to Visual Studio 2008.

The story will continue, no doubt, for several more years. I am still developing most everything in Win32 Delphi--as a vast number of Delphi developers are. I suppose someday, Microsoft will force the issue and stop supporting the platform, but when I look at the applications stil coming from them, I'm not worried that it will be any time soon.

The .NET platform is well-engineered and has lots of features built-in that currently require 3rd-party libraries or lots of programming for equivalent functionality in Win32. I am excited about learning the new technology, but there is always a learning curve and there must be a good reason to move.

For now, I will continue to support and trust the Delphi product line. I still believe it is a very solid product and I also believe there will be a company behind it for quite some time.

Add new comment

The content of this field is kept private and will not be shown publicly.