Spring 2008 Excelsior JET Customer Survey Is Over
Share:
The customer survey we have been running since the end of February is now over, and I want to share some results with you.
But, first of all, our congratulations and the $100 Amazon certificate go to our prize draw winner, and a big thank you goes to all the participants. The response rate was overwhelming: over 32% of sent invitations resulted in a filled survey. Thank you very much for your feedback, we appreciate it!
(The prize draw winner is a person from Brazil whose company has purchased Excelsior JET Professional Edition for Windows in mid-October 2007. We have not heard back from him yet, so I do not have permission to disclose the names. Notice to the winner: If you are reading this, please get in touch with us at earliest convenience, or we would have to choose again.)
Back to the survey, let me begin with the Java IDE usage profile. It turned out that over 65% of Excelsior JET customers use Eclipse, and about 25% use NetBeans, the latter being a bit surprising for us. Our favorite IDE, IntelliJ IDEA, occupies the third place with 8.7%, followed closely by JBuilder (7.7%). A few people use JCreator, while each of CodeGuide, EditPlus, Metrowerks Codewarrior, SlickEdit, Textpad, and XCode were mentioned only once, as well as Oracle JDeveloper and IBM WebSphere/Rational tools. One person uses MS Visual Studio 2005 as a Java IDE, which must be an interesting experience. Finally, 11.5% of all respondents indicated that they use no (Java) IDE at all. So do I – hello guys!
(For the sharp-eyed reader: yes, it was possible to specify more than one IDE, so the total exceeds 100% with about 1.37 Java IDEs per Excelsior JET user 
)
Every second respondent has expressed interest in sans-install distribution. It seems however that in most cases the Self-Contained Directory option, found on the page Backend in JetPackII, covers their needs.
Over half of all respondents use Ant to automate their builds, but most of them find calling jc and xpack from their Ant scripts good enough. Just about 5% of Ant users have expressed interest in tighter integration.
Now, the most interesting question that we used as the headliner: Excelsior JET for Mac OS X. First, it turned out that 60% of existing Excelsior JET users are not interested in us adding Mac OS X support. Then, only 12 people said they would pay for the Mac OS X version in advance. Unfortunately, even if all of them would agree to pay, say, the list price of Excelsior JET Pro for Windows in US Dollars, that would allow us to accumulate only $27,600. That is a nice sum on its own, but it won’t help us much, as our rough cost estimate for the Mac OS X port is in the low six-figure range.
Still, 40% of existing customers expressing interest in a new product is a good motivation to seek other sources of funding. Compare this to the 32-bit Solaris port: less than 11% of respondents have expressed interest, and nobody is willing to preorder it.
Note: Comments are now closed to avoid spam, but if you have something to add or ask about, please contact us.
April 30th, 2008 at 8:40 pm
I understand that Excelsior is working also on a 64-bit JET. Please keep us updated.
April 30th, 2008 at 9:16 pm
Once you have a working version of the compiler in Linux, how difficult is it to port this to Macintosh? Could I compile with the Linux compiler and run the code on a (x86) Macintosh already?
Thanks!
James
May 30th, 2008 at 3:59 am
It’s kinda strange to use the opinion of the existing WIndows-Unix customers to reason about Mac OS version.
If you guys can deal with SWT Cocoa and RCP stuff in a smart way, you’ll hit Mac OS market.
Btw, Java for IPhone is to be released this summer. Thus, a lot of developers will have an urgent need for slimming down their apps to deploy them on IPhone.
Pls, think about this.
May 30th, 2008 at 12:34 pm
James, the compiler and the runtime are easy to port as Macs are now x86-based and Mac OS X has BSD Unix as its core. The problem is the standard library, most notably AWT. We have a license for the Sun’s reference implementation of Java SE, so we just use their code in Windows and Linux version of Excelsior JET. But Apple did the port themselves, and they won’t let us use their code. And the Windows AWT.DLL is over 900KB in size…
To put the long story short, the Mac port would be at least an order of magnitude more expensive than the Linux port we’ve done back in 2004.
May 30th, 2008 at 12:43 pm
Maksim, we have a very good reason to rely on paying customers’ opinion much more than on the opinion of would-be customers.
Many years ago, we were selling a much simpler and much more portable product – command-line Modula-2/Oberon-2 “via C” compiler. (It is now known as XDS-C and is available as freeware.) We had versions for Windows, Linux, Sun OS, even Ultrix. And we have been receiving inquiries like “I would buy your product if you port it to platform X.” So we did the port to that platform and did not sell a single copy.
Now, you know what? That “platform X” was Apple Macintosh.
Considering also my previous comment, we may not afford the risk on our own at this time, and there are not enough people willing to share the risk.
May 30th, 2008 at 6:24 pm
Dmitry,
I am in no way challenging your business intuition and market expertise.
Personally, I am now looking for oppotunities to optimize performance and deployment cost of OSGI-bundles on Mac OS X (and I don’t care about awt). However I have no data if there is a critical mass of guys like me to allow you pay the bills.
However all these things considered, I am not sure that Oberon/Modula compiler many years ago can form a solid ground for marketing analysis of Java AOT compiler today.
Furthermore (just a shameless dream) since reimplementing awt is very painful, maybe you can consider some flexible offering that allows to deploy certain stuff and yet not bearing the label “100% fully certified pure java”.
May 30th, 2008 at 6:42 pm
Maksim,
I am sure there will be customers for the Mac OS X version of Excelsior JET. The point I was trying to make is that not all potential customers who say they would buy product X would actually do that when product X becomes available.
Of course, if they pay in advance, or agree to share the risks and profits, that’s a totally different story. Perhaps we could try to set up a foundation of some sort.
We may not sell a product that has not passed the TCK. There is an option to pass the TCK in so called “headless mode, that is, without display, keyboard, mouse or sound. So we could port the core and let you rely on SWT for GUI. However, all people who have expressed interest in the Mac port up to this day were using Swing…