Kylix from Inside (by Michael Swindell)

In a set of newsgroup posts, DevCo Product Manager Michael Swindell tells us a lot about the internals of the Kylix project.

In a thread on borland.public.delphi.non-technical asking for a Kylix survey, Michael Swindell made several posts revealing many interesting and unknown facts about the Kylix project. Here is a summary of the key points he mentions (emphasis is mine!):

"Much analysis and debate was had over GTK vs QT as the underlying drawing and widget lib because unlike Windows, Linux had/has a fractured UI community and there certainly wasn't a GDI or native controls on Linux, unless you used Wine which we decided to use for porting parts of our IDE, but not for development or end user apps. There are good reasons for going either way, in the end we chose QT because we liked where KDE was going and felt it was closer to the Win32/GDI direction and therefore cleaner to wire up to VCL."
"From certain standpoints, Kylix did ok - from many external perspectives it would have been considered a blockbuster success, from other perspectives it never came close to living up to its promise or potential. When it launched I think it was selling more in license $ (not in support/services) than any other Linux application, tool, or distro that year. That was the year Borland's stock briefly hit 20+ for while on the Linux boom. I like to think that Kylix generated moderate millions in sales, but a half billion dollars in Borland shareholder value -- albeit briefly."
"It needed to sell several times as much in order for it's existance to meet the business criteria we were working within... And then to get increased investment it needed to be growing. The other pressure was Delphi, for the most part we had one team so if we were spending a year doing a Kylix release, that was a year that we did very little on Delphi, which was our core business in the RAD group... For Kylix to have continued it needed to perform a lot better than it was, it just wasn't coming close to where it needed to be and we could not afford to starve Delphi."
"The hybrid Wine ported IDE was also problematic for us. The R&D team did a great job, but it was a technical minefield - it's not exactly your everyday ordinary app, things are going in and out of different processes left and right, debugging, designing, ported stuff, pure native stuff, it was extremely complex and Wine was also not as mature then so we were doing a lot of contributions and working closely with the Wine guys. If we were to start today it's likely it would be built on Eclipse, MonoDevelop, or KDevelop or something else... but probably not on a Delphi Windows IDE port."
"For the new company we will be bringing all of the Kylix intellectual property with us, so we will certainly take more looks at it as an opportunity. When we first developed Kylix we intended at some point to create a cross-compiling and cross-debugging Kylix plug-in into Delphi - similar to Simon Kissel's CrossKylix (great stuff!). That might be an option someday, but we won't be making any decisions on Kylix until after we're spun off - today the product remains frozen and our focus is on the Windows platform."
"Regardless of what we do from a product or development perspective we've made some past commitments to the community in around Kylix technologies and open source that never came to fruition and I hope that we're able to make some of the promises finally bear fruit."
"For Kylix to work in the future, it really would need a different cost structure, and also some strategic adjustments - ie where Kylix shines, is not exactly where the Linux market has grown and done best. Anyhow, as our own company we'll certainly creatively evaluate and reevaluate the opportunities for Kylix development. Many of us in DTG worked on it and were very proud of the product and technology."

What should I add? Not much. Some of this is really interesting info, and it you read the entire thread you'll also find an analysis of the money the product should make in comparison with the R&D costs to be successfull. The only element I want to underline is that when Borland announced the Kylix Community Project I volunteered to be part of it. Nothing came out of it, but I'm still here, still using Kylix, and certainly interested in helping DevCo with its Kylix plans. Just give me a ring!

Why one of the ads by google on this blog links to "Endangered species"? Yup! Is Delphi now protected by WWF? :-) Swindell post is interesting, it was a pity they did not choose the cross-compiler then. It would have required less resources, and would have let them build a native IDE later, if required. But if they think they can build one on Eclipse or other tools, well, it looks like DevCo. tool is surrendering really as an IDE builder. BDS has already troubled its users enough with its VS-like features.

As always: "...until after we're spun off ..." "For the new company we will ..." Heck, they have to get the spin off finalized! It's about 9 months since it was announced. It's not product strategies ruining DevCo - it's the lawyers! Regards, Olaf

This confirms what I suspected in some ways. Namely that Borland placed full commercial revenue expectations around a product that would be required to succeed in the Open Source space. This was and is unrealistic. This also underscores another point I've made previously on other blogs; Borland isn't looking to develop a market, lead in a particular product space, nor innovate in non-traditional ways. One of the core REQUIREMENTS for success in the Linux/Open Source space is CREATIVITY. Another, which goes hand in hand with creativity - also known as INNOVATION - is COMMITMENT. If Linus Torvalds, who by the way does pretty well financially, certainly better than any Borland 'employeez' - was only looking to 'make a fast buck' with his Linux project, we wouldn't have Linux today. For certainly as a new kid on the unix block, he could never have competed with the R&D budgets of Sun, HP, SCO, IBM and others. Instead Linus' path to success with the Linux OS would require a long term commitment to producing a flexible, dynamic, high quality, high performance product that could positively impact the 'small' computing world, i386 etc. In other words, his efforts had to be, by necessity, a 100% risk venture. Therefore it had to be fun, challenging, and provide it's own reward regardless of any hope of future revenues that it could generate. The fact that Kylix was profitable, demonstrated to the business community that Borland had a solid commitment to the future of computing (read LINUX). This was reflected in the stock price jump. It appeared then that finally, Borland appeared to be on track to outgrow the stigmas that the brand carried with the Lotus lawsuit and the years of horribly incompetent corporate product management. Years ago, having been active in equities trading and software development for the same I was able to demonstrate using historical charts and graphs that the time of greatest potential for growth and success in Borland's financial future was when it appeared that Borland had finally graduated to the 'big' leagues (read unix) and was going to continue it's previous market leading position in s/w development technology innovation. This particular market is not the typical 'Windoze weanie', flash-in-the-pan, fly-by-night technology market. You know what I mean; throw some nice catchy marketing buzzwords out there, add a few features and tweaks to the same old product mix, choreograph a new box label, and sell a few million copies on marketing hype alone. No the unix community is where the windoze nonsense hits the wall; here is a product space where companies typically would spend significantly larger sums of money for their development products, and expect significantly higher quality product/support mixtures. Linux was and is an excellent opportunity for Borland/DevCo to do far more than make a fast buck. This is the kind of space that innovative, committed geniuses like Larry Ellison of Oracle $DEFINE. This is the Wild Wild West of computing; the early entries that buy the real estate at dirt cheap prices are the ones who get to define the future of the real estate market. This space is currently CRYING OUT for good dev tools. If you want to make the big bucks, you have to have BIG COMMITMENT. This means being committed to $DEFINING the dev tools market on Linux, not trying to sell a few million copies of your product on successive hype cycles alone. I was once priviledged to work on a system upgrade project for a large bank that has over 6k branches in the US alone. They traditionally had used a windoze desktop/legacy unix server architecure to tie all of the data collection activity together into one humongous real-time data processing powerhouse. They built their success on the solid, stable, high performance platform of unix computing & internetworking to build one of the largest retail banking systems in the world - in the top 10, I believe. On one of their previous system-wide upgrade projects they intended to replace all of their legacy unix deployed servers with windoze terminal servers. Try to imagine the cost and complexity of upgrading over 6k branch nodes of your national WAN with a new terminal technology; you can easily experience having your eyes roll back in your head at the sheer logistics of it all. To make a long story short, they had to roll back to the legacy unix systems because of the instability/lack of quality of the windoze systems. They were completely unprepared for the increase in cost/downtime associated with the lower quality windoze product line. They had to learn the hard way that quality doesn't happen overnight. On the project that I was working on I acted in support of the field deployment force for new desktop computers. I had a field installer call me from one state with a SEVERE problem; the keyboard and monitor no longer worked on the unix terminal server; the installer felt that his installation for that site was a total loss. With great deliberation I asked the installer if he had hooked up and imaged any of the new windoze boxes on the wire network, he said that he had. I instructed him to hit the start button, choose 'Run', and type 'Telnet servername:port' and enter the server's login credentials. The installer was able to complete the installation without further assistance. To wrap it up, this kind of quality requires the kind of commitment that makes great software great and great companies a buzzword in the corporate world. This hit and miss nonsense of always trying to guage the 'optimal entry point for the current market trend of a specific product space' makes the company thrash around like a whipsaw; it's a great way to exhaust your best talent and achieve high turnover in a fickle market. It also makes your company appear unstable to the business world and a bad investment risk. Once again I implore the folks at DevCo: STOP IT! STOP IT! LEAD! Stop following the 'herd' and get back to your only source of continued success: Creativity/Innovation/Commitment. Learn the lesson of Phillipe Kahn: find the opportunity, design unique offerings, and work hard to make it a success. Build an organization that can carve out new product space and be mature enough to drive it and support it.

I have always believed that Borland could do much better if they picked up a linux distro and modified and made a tool for that distro. I even wrote to Borland 7-8 years back about that. I think Borland should pick a distro and just make Kylix work with that distro. There are so many Linux distro's and I think it might a bit difficult to make Kylix work with them all. Sandeep

To summarize, they tried to do it fast instead of right and that it was both a technical and financial failure from their point of view. None of this is a surprise to anyone that who looked at the whole project without putting on the rose colors linux blinders. Will we see Kylix rise again? No. The comment about the IP going with them? Not news, it is worthless to Borland, OBVIOUSLY it is going with them. Will they look at a cross compiler & debugger? No. It could easily take a year to get that working CORRECTLY instead of quickly and the spin off simply will not have the time or resources for that kind of effort for years. Make no mistake, is spite of the recent improvements, Delphi is still a train wreck. DTG is going to have to spend a HUGE amount of time and resources not only getting Delphi & Delphi.Net back on track and up to date (here's hoping the next roadmap isn't as bad as the last one or they are dead in 5 years regardless), but now they have plans to invest a large effort into making C++ work properly (not just work, but actually be worth paying for and using in commercial products, not the pushed out the door half baked offering they are still patching). Nope, you can be sure that there will be no resources to waste on a market that they could not properly monitize the last time. DTG claims they'll be in control and get to call the punches. They are VERY mistaken. Investors will only let them go so far before they are reigned in and have to justify how the money is getting spent. Based on Kylix's past failures and state of the product, that would be an EXCEPTIONALLY hard sell. Nope, when it comes to Kylix, don't hold your breath.

Marco Tech Blog