Slashdot Mirror
Inprise/Borland Developers Conference Linux Nuggets
Inprise has offered Interbase for Linux for over a year now, which is very much worth looking at--but the problem is, the Linux version is only 5.1, where the NT, Solaris and HP/UX versions are all at 5.5, so there's been some question about Inprise's commitment to Linux. At the conference, they announced that Linux development is still going strong, and that while Interbase 6.0 would be released for NT first, its Linux release would be in the following quarter. The NT release is currently planned for mid-2000.
Inprise also announced JBuilder for Solaris and JBuilder for Linux. This is based on what they call "JBuilder PrimeTime" which is JBuilder in Java. Apparently JBuilder was originally written in Delphi, and over time they have been porting pieces of JBuilder to run natively in Java. The currently shipping JBuilder 3, which is only for Windows, is not yet 100% Java; however, JBuilder for Solaris, which will ship by the end of the year, will be. It will be followed up in short order by JBuilder for Linux in early 2000. There are strong rumors that JBuilder for Solaris was moved up at Sun's request, because they want a better development tool on Solaris than Java Workshop to prop up their hardware sales in the developer market.
With all this Linux stuff going on, there were a dozen or more Linux-oriented sessions at the conference; quite a change from last year. And of course, the Delphi team had to get their hand in. They did not make any announcements; in fact, they made it quite clear that they have not yet decided what to do. However, they did give out some results from the developer survey: The most interest is in Delphi, followed closely by C++Builder; the vast majority want a full RAD environment, not just traditional tools; native GUI support is by far the winner (as opposed to Motif or [shudder] WINE), with KDE preferred over Gnome by a small margin--though it goes unsaid that if you could have chosen more than one, everybody would have picked both. They have had 16,000 responses since the survey went on-line, which they consider an excellent turnout.
Following these announcements, they did a demo of what they've been working on so far. As many of you know, the back-end compiler is the same between Delphi and C++Builder, as of Delphi 3/C++Builder 2. What they showed was bcc running on Linux (specifically, Red Hat 6.0 using Gnome). They compiled Xgalaga with both gcc and bcc, with bcc getting the job done in a little less than half the time that gcc took. They then ran the resulting binary to show that it was real, working code. Watching the demo closely, it was also possible to see that the binary that resulted from bcc was smaller than the one gcc produced, though there's no way of telling whether this was due to smaller code or just different debugger settings. And of course the real interest is in the RAD environment, which they did not demo; the question of whether Delphi for Linux will include the VCL is yet to be answered. But it's clear that work is ongoing, and if they've already invested enough time to get bcc running, it seems quite likely that a product release will actually happen someday.
Of course, Inprise/Borland is not an open source company and all these products will be commercialware, and there are still issues to be resolved concerning how quickly they will support new kernels, libcs and distros; how nicely they will play with the other kids; how the community will react to the concept of for-pay, closed-source development tools; whether their licensing will permit you to build open source projects using Borland tools; etc; etc. There are many ways they could screw things up, and even if they do everything right, it is not at all clear that a large market exists for for-pay development tools, in the presence of really quite good free ones. However, I think the "linux market" and the "linux community" are increasingly divergent concepts, and there may well be room for Delphi somewhere in the market--if perhaps not so much in the community itself. "
First - the legalese : I'm an employee of Borland / Inprise but I'm writing this in strictly personal and entirely unoffical capacity. Everything I say here is completely devoid of any offical status.
That said;-
Yes, we are serious about Linux. We're a commercial company so we have to find some way to coexist with Open Source to our mutual benifit.
We think we can do that and offer useful products to Linux developers and commercial users of the OS.
As to the products; I have the Java release of JBuilder running on my Linux machine right now; and it is awesome
I also have the tech-demo of IBM's Linux Visual Age for Java and that, too, is excellent. Which one you prefer comes down to the RAD paradigm you like most.
Linux JBuilder needs JDK1.2 to run. VAJ is (mainly) native binaries but the two IDEs seem to run at the same speed. The JBuilder team did an excellent job of getting JB to run very fast and stable.
As noted, our port of bcc to Linux compiles about twice as fast as does gcc. I do not have direct experience of the compilation times at present but I am informed that it does indeed run faster and produces smaller binaries. The exact gains depend on the compiler options selected.
We also have a bunch of other stuff poised in the wings awaiting offical release...
Cheers,
Linc.
Overall, I think the Microsoft Developer Conference was a bit more lively, though. (:
In Unix, bcc is a GPL'ed compiler for producing 386 binaries. I realize that bcc has been the 'Borland C Compiler', but we really don't need more than one C Compiler on the same platform, with the same name. That's just trouble waiting to happen.
pb Reply or e-mail; don't vaguely moderate.
The only way would be to find the best possable flags for that specific system, for that specific code (in the above case, a benchmark itself written in C). And the results would still be suspect because it would only show the results for that specific hardware, and that specific code.
In my mind, my solution is this, if/when I want a new compiler, I realize that I have got by with gcc/egcs for such a long time, I'll find the person who is working on the code specific to the hardware I am running, and donate money or equiptment directly to that person. Only other choice would be to pay for something that is "closed" and hope they did a good job, and hope it was a value. ;-) So...
gcc/egcs isn't perfect, and has a ways to go, but it sure is a dang good value, and I believe it will get better. When open source comes up, I belive gcc/egcs is a better success story than Linux, FreeBSD, or any other piece of software, because without the compiler, the software would never be were it is today.
I'm stuck writing code for NT at work. I'm working on some Linux code but it's not officially sanctioned. I'm doing it to prove to management it can be done and work as well if not better then NT.
Most of the code is done in C++ Builder because the GUI development is extremely fast and flexible. If Borland would release source compatible VCL and C++ Builder for Linux I don't think my management could argue againts doing everything in Linux. On top of that I could just re-compile all my apps to Linux in not time.
Ex-Nt-User
I wonder if it will compile the kernel. I am curious as to the difference in size and speed of a gcc-compiled and bcc-compiled linux kernel.
Is the kernel gcc-specific?
It would be interesting to try on things like X11 and window managers; things of that sort.
-Gabriel
Nice story. Had me fooled for a few seconds.
D
----
Relax, everyone knows that AOL ripped off IBM's innovative OS/2 SHIFT+INSERT technology, which failed only due to POOR IBM MARKETING!!!!
I'm sure Apple is somehow to blame too.
--
Business. Numbers. Money. People. Computer World.
I noticed one of the options on the developer survery asked whether there was any preference for using existing open source tools.
IMHO it would be very nice if Borland/Inprise used GCC and the other GNU tools as a backend (and submitted improvements to their code bases as needed!) while using their IDE and RAD tools on top. I think people would pay for this, the community would benefit from some source improvements, and there would not be the overhead of supporting another compiler. Metrowerks seems to be doing this with their Codewarrior "GNU edition." Does anyone know how well this works?
Borland would have to be careful not to violate the open source tool's licensing, but this shouldn't be impossible.
-OT
As for GUI development, VC++ is really not any good in that environment. After having use Powerbuilder and (shudder) Visual Basic, I can't see any reason at all to use VC++ for GUI development. And now you have an interface that mimics VC++?? For shame!! Everyone is entitled to use what pleases them, of courses, which is one of the guiding principles behind Linux, but what we need is not imitations of life-sucking programs which Microsoft foists on the public, but innovative new programs which foster RAD, or for companies to port existing excellent software over to Linux, like Inprise is doing.
That being said, let me qualify by saying that I've never tried Kdevelop, and probably won't ever, because it requires Qt/KDE, and I have enough libraries without adding another set + desktop environment.
Still, anything that reminds me of Microsoft on my Linux machine is summarily deleted.
Matthew Vanecek For 93 million miles, there is nothing between the sun and my shadow except me. I'm always getting i
Windows-style development tools allow both the API designers and the application programmers to delay facing up to, and fixing, serious design problems with their systems.
For example, the MFC class libraries have serious problems in their inheritance hierarchy, but rather than fixing them, Microsoft just adds a bunch of wizards to their development tool to do manual delegation.
Similarly, the VC++ GUI design tools result in GUIs that, for the most part, use fixed layouts that don't work well at different resolutions, and they generate code that is difficult to maintain.
To me, Windows-style development tools give the appearance of making development easier, but in the long run they have a negative impact on the quality of both the OS and the application code.
Bringing a lot of the Windows development tools to Windows risks bringing Linux down to the same level as Windows. I hope that the traditional Linux community will be strong enough to keep those kinds of influences from doing real damage.
Linux development tools aren't perfect, of course, but I think Windows-style development tools are the wrong place to look for improvements. We don't need more windows or buttons, we need better tools support for specifying, composing, analyzing, and verifying large software systems.
(In case you are wondering, yes, I have developed in VC++, Borland, and UNIX, and I'm currently a full MSDN subscriber.)
Has anyone tried FreeBuilder (www.freebuilder.org)? How does it compare to JBuilder? It's supposed to be pure Java, and running on Linux now...
-- What you do today will cost you a day of your life.
Anybody else notice that this is the first wide-spread survey regarding developer mindshare of GNOME and KDE? Seems they're about even, which I suppose is a good thing. Sure, the statistics are probably biased (maybe one group hates Inprise or commercial software more than the other; maybe one group hates surveys), but it is neat to see.
Borland has a strong compiler, which has always done well in performance, both compile time and run time. It has also done well in ANSI compliance, and as it has been the common back end for Delphi and C++ Builder, Borland's commitment to making it good is the at the core of their business.
Borland also has, in Delphi and C++ Builder, strong RAD tools which will make it very easy for thousands of Windows programmers to move apps to Linux, furthering the growth of the Linux community.
The range of their questions in the survey, and their willingness to publish preliminary findings shows that they are committed to delivering what the market seeks.
As to commercial software, I welcome paying for a great tool, rather than getting a good one for "free". My time is worth much more than I will pay for the tool, and my experience with Borland tools gives me confidence that lack of source access is not an issue.
Borland's VCL has not been open source, though developers do have source access to most of it. There are patent issues involved, and I would expect to see the source access to the VCL remain much the same under Linux.
I have already invested thousands of dollars of my time in Linux, and am not yet ready to try to ship a product. If I had Delphi, or C++ Builder, I would have reduced that investment by an order of magnitude, and would be hard at work on my applications.
Please, Borland, move with all possible speed!
--- Bill
There is one thing that your report does not mention about the gcc vs. bcc comparison: was there any way to compare the speed of the compiled code?
It is nice to know that bcc compiles faster than gcc. This is interesting for those of us who are writing code and compiling all day. But when the end users install the software on their machines, they mostly care about the speed of the executable (compilation time does not matter much).
-Raphaël
I don't see a huge difference between the linux market and the linux community. They are mostly one and the same. There will be some people who aren't in the market, refusing to buy anything, and downloading everything. These people are rare (even 1.89$ for a distro disk counts as a purchase). Then there will be the truly clueless who buy JBuilder for Linux to run on their WinBox. Everyone else is both using Linux and buying stuff for Linux.
Yes, there are hardcore Linux users and lightweight, parttime Linux users. But they are both part of the community.
A Government Is a Body of People, Usually Notably Ungoverned
There's a lot of issues that come up when a commericial compiler will be available that is likely to become very popular. I'm willing to bet that lots of people would buy Delphi for Linux (including myself). But, I want the apps that I write to be runnable on alphas, sparcs, ppc, as well as x86. Normally, with gcc, this isn't a problem -- you simply distribute the source with your app and let them compile it... or get a friend to do it if you're weird. There's one thing that you CAN assume -- that almost every Linux user will be able to use your app... because almost everyone has gcc.
Now, with Delphi I don't think you could do this. I don't know how I'd be able to ensure that Linux/Alpha users would be able to use something I programmed in Delphi. So, I came up with one idea.
What if Borland freely distributed their compiler that handled Delphi source code, for all Linux platforms? Maybe they're already planning on doing this, but I'm not sure. Anyways, they could still sell the RAD environment and the tools. This way, you could even write open source Delphi applications.
One problem with this is that people would be able to code without the RAD environment, and just edit the source files.. and thus not needing to buy Delphi. But I'm sure Borland could figure a way around this.
So, to conclude, I guess what I'm trying to say is that Borland should seriously consider the ability to produce applications that can be ran on all Linux platforms, not just x86 / glibc2 / qt / whatever.
I really wish the guys at Troll-Tech could work out a licensing agreement to make a version of Qt price-competitive with commercial Motif. I mean a Unix-only version exactly like Qt-free, but with a license to let you produce commercial software with it. It's an amazing product, and we could see Borland QtDelphi and C++-builder with Qt-GUI. Mmm...
Commercial and free development can, indeed must, exist side-by-side.
BTW: Whoever mentioned that they like Visual C++ should check out KDevelop (www.kdevelop.org). Its interface is basically identical to VC++.
--JZ
um, you did you know you can get rid of all that crome, right? all of those toolbars can be removed, and the individual buttons can be put wherever you want. so you can make one toolbar just for yourself.
I've removed everything but the menubar (and I put 3 buttons on it, fullscreen, save all, and the button that shows/hides the output screen).
but I can understand why you wouldn't know that, it's so hard *right click in the work space* and all.
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
ReadThe ReflectionEngine, a cyberpunk style n
VC++ has a nice interface ?
The Borland C++ 5.02 (I use it in college) is so much better - Clean , uncluttered and only a minimal interface.
VC++ has to put everything on the screen, only a quarter of my screen is usually available, half at max.
I can throw myself at the ground, and miss.
That wasn't demonstrated. In fact, showing bcc was just a teaser to whet our appetites for anything Linux. But, both executables ran identically.
As for the "Primetime" (JBuilder in pure Java). There were quite a few discussions and demonstrations. The one thing they did demonstrate was running the same Primetime for Solaris under Linux. What's holding back Primetime for Linux is that the Blackdown JDK still has too many bugs (heck's that's why it's prerelease).
I asked about whether Borland was contributing to the Blackdown project and was assured that they were. Also, Borland demonstrated their Java JIT compiler. This made Primetime perform like a native binary rather than an VM'd app.
Borland is also said convincingly that they will not develop their own VM. Instead, they said they will develop for the one in most common usaga on the Linux platform - Blackdown.
Among the other things demonstrated were:
DataStore - An object oriented persistent model for Java.
Delphi CORBA going against a Linux based CORBA server. They also demonstrated a preliminary IDL compiler that worked with Delphi code.
The commitment from Borland appears to be there. It won't be long before their tools will be among those of choice -- especially in a corporate setting.