Slashdot Mirror
Borland/Inprise Linux Survey Results
Ronin Developer writes "Borland has just released the results of their Linux survey. The results are somewhat interesting. Check it out for yourself at:
Borland Linux Survey " There's a lot of stuff there about
what Linux users say they want (in general, and from Borland)
What surprises me is that 26.8% of the respondants have never heard of GTK+, Qt, Motif, or Swing, or they don't even know what toolkit they want to use.
Who the heck is responding to this poll?
You could answer multiple choices on those questions..
-- I'm the root of all that's evil, but you can call me cookie..
Seems most are interested in developing for Linux using RAD.. The prefered RAD tool tends to be Delphi for those responding.. 'nough said.. ;-P
-- I'm the root of all that's evil, but you can call me cookie..
I've been working in Delphi for two years, and I've got to say, I love it. Before that, I used Ada, C, Perl, TCL etc all on Unix (Solaris & Linux), and I can safely say I can produce working, stable code 3-4 times as quickly in Delphi as in anything else I've ever used (that includes some pretty expensive Unix tools like Rational Apex)
If there was Delphi on Linux now, I know three major ($300 Million+ Australian turnover p.a) companies (that I've worked at) that would start coding on it by the end of the year - and this is for Client side apps, not server applications.
FLAMEBAIT
About the survey, I think the KDE/GTK+ thing may have been that there are many people using the KDE now, who would prefer to go to GNOME & GTK, but are waiting for more stability.
/FLAMEBAIT
Also, whoever said that the an option in the survey was to have MS SQL Server on Linux was wrong. The question asked what databases you would develop for. I know we are developing for SQL Server, and I don't think that's going to cahnge as quickly as swapping to Linux for the client apps might. On the other hand, I've developed Delphi stuff with Oracle Databases on Sun E1000 (or whatever they are called), and they were looking at other options.
Well, I've never had to program in VB, and I can't say that I truly dislike Delphi, but at least to me, it is still Object Based, and not Object Oriented environment.
When I look at Object Pascal, I see many of the same mistakes in the language that are present in C++, and which I think in some cases break the OO paradigm and add unnecessary work, or bookkeeping, to the shoulders of a over-stressed programmer. In my book, Object Pascal is a hybrid language, or Object Based, as you put it, and in the same category with C++. Unfortunately, I think hybrid languages have little use. If I need to do quick and dirty, efficient code, I'd rather use C, if I need to add to the abstraction level, I rather go for languages with much better OO support, such as Java, or Smalltalk.
Another thing that bothers me with Delphi is the way the whole project is tied to its IDE. My first experiences with Delphi was about 6 months ago, when a project was handed to me which uses Delphi to build its GUI (the layers underneath are pure C). I found it extremely difficult to understand the flow of the program, when I constantly had to switch between looking at the code editor and the object inspector to figure out what the heck was going on with the program flow. Very frustrating. Another thing I think is a bad idea is this whole aspect of building a GUI first, and then attaching bits and pieces of code scattered around the GUI components to do the work. It's ok for quick prototyping but ends to lead to a bad design if used to build the entire project. You can avoid this, of course, to an extent, but the Delphi IDE is build in a way that makes this harder than doing it the Delphi way, which means most programmers just won't bother fighting the IDE for this.
Anyway, there are alot of loyal, and quite vocal supporters of Delphi, and I listened to them and had high expectations when I first had the chance to start learning it, but I was somewhat disappointed when I actually got to see what it was.
Oh well, that's life I guess..
Excellent language for doing what? To me, Object Pascal looks like a copy of C++, with most of the same mistakes in it. It still breaks the encapsulation of OO, at least on logical level, as C++ does. It's a hybrid language, trying to be both procedural and OO at the same time, as C++ does.
Since Delphi seems to be the only environment where Object Pascal is used, why not make it a truly Object Oriented programming language, and loose the procedural stuff in it, thus gaining the benefits that OO paradigm promises to deliver.
Contact Halycon
http://www.halycon.com
Hetz (Heunique)
The copy of C++Builder 3 (+later pro.) I got with PC Plus both worked fine (although I'm limited by the license).
The reason for the "finally" clause is that it is very convenient. There are often reasons to do something at the end of a block without deallocating anything. I may want to make sure a table gets closed without deallocating the table, for instance. The "finally" clause has been part of the try block since that structure was invented, long before Object Pascal.
Pascal does goof though in that it will not let you use a combined block:
try ... except ... finally ... end
instead you have to write
try ... try ... except ... end ... finally ... end
I couldn't possibly disagree more, especially the last part. Qt is a separate, distinct GUI toolkit. I'm using Qt in my latest software project, and I have no plans of developing for KDE in the near future. I'm sure I'm not alone. There are many reason I, in general, prefer Qt over GTK: C++ vs. C is one - it always frustrates me when people use the fact that something is written in C++ as a negative; I'll take elegant, well structured, true OO C++ over plain C for something like a GUI toolkit anyday of the week. GTK is OO, but doing OO in plain C is, IMHO, just plain silly when you have C++ (if the platforms you're targeting don't have a decent C++ compiler, that's another matter). C++ certainly has more (way more) than its fair share of faults as a language, but in the hands of a top-notch designer who truly understands OO and the intricacies of C++, it gets the job done and it gets it done efficiently. This whole business about C necessarily being more "lightweight" than C++ is hogwash. If you know what you're doing, you can often make C++ code more efficient than C. A lot of people will look at me funny when I say this...ultimately, anything you can do in C++ you can do in C...you just have to work a lot harder in C to do some OO things. But that's exactly the point; if you're programming in plain C, you'll often not use a more elegant OO technique (which may ultimately be more efficient) because it's so much more work. A C++ programmer is free to use polymorphism, et al, where it's appropriate without jumping through hoops. The problem is, there are a lot of programmers out there who really don't know C++ as well as they should, and who aren't going to be able to pull off a truly elegant design. This brings me to my next reason for prefering Qt; it has a very elegant design. Those guys at trolltech know how to execute an OO design in C++. Let the well trained OO designers handle writing elegant OO toolkits (like Qt); the masses (many of whom aren't nearly as well versed in good OO design) use the toolkit and its structure to lead them to a good design.
But the reason I chose Qt over GTK is very simple...I needed an elegant GUI toolkit for Win32 initially. I wanted one that also supported X11/*nix. Qt is very well supported on Win32. The Win32/GTK+ is pretty sketchy right now.
>I am sure that anyone who is interested in developing object oriented code who's looked at Qt will instantly prefer it over GTK+. amen
Think about the people who use Delphi and want to move to Linux... They're that 95% (or is it 90?)
Well, I'm one of them. But I don't need it at any cost, neither quantitative, nor qualitative. They want a simple way to develop software quickly, and that means RAD tools and big libraries.
Yup, I've been there too. 'We need it yesterday, man! Move!' Usually it ends up in fighting fires the last night before deadline and of course it doesn't help and finally the teams needs two more months to fix it.
KuroiNeko
Object Pascal by far more suited to RAD than C/C++.
:-)
Pascal has strong type checking, and the structure of the language basically FORCES you to write good code. C/C++ is a good choice for professional/sem-professional developers that need a good, robust language.
But for a one-man programming project, particularly when RAD methods are to be used, Delphi and Object Pascal puts C++ to waste.
Most Linux programmers, OTOH, are experienced hackers that think that definition of "Languages of Choice" certainly does not include Pascal. In hacker-style coding, C++ is certainly preferable, but for mission-critical code that must be developed in a hurry, one certainly cannot beat Delphi. The only other choice for RAD is Visual Basic, and we certainly won't see Linux programmers flocking to that any time soon, for obvious reasons.
My journal has hot
If they just could do the same and also bring a commercial compiler to the linuxppc communities and other supported processor by Mot/Metro (ie Coldfire 68000) and also the ARM. ....
Gcc is prety cool but I know commercial compiler are better in code generation than gcc is
none Yet.
If you want a taste of the old days, they DID open up the old Turbo C up to version 3.0 for FREE.. ;-P
-- I'm the root of all that's evil, but you can call me cookie..
The answers here don't suprise me. If you've decided to use a widget set and not a whole complex environment, the odds are you've decided to make things more lightweight. Well, GTK is just plain C under the hood, and Qt is C++ under the hood. I don't think many people would disagree with the assertion that GTK is more lightweight than Qt. If you've decided to use a whole environment, you're more likely to be doing really complicated things in C++. Given that, there's less reason to select GNOME instead of KDE. Also, I can easily believe that there are people interested in using GTK without GNOME, but fewer people interested in using Qt without KDE. So, the results seem perfectly consistent to me.
I don't know how much this poll can be trusted since it was linked off of one of KDE's pages and they encouraged everyone to go there and fill it out. That's how I found the survey anyway.
I originally found this poll linked on one of KDE's pages. Unfortunately, I don't think the results of the KDE/Qt vs GTK+/GNOME polls can be considered accurate because of this.
What does being from Europe have to do with using Delphi? I'm from Europe, and even I was surprised by the results that Delphi got. Of all the programmers I know (quote a lot) only very few use Delphi.
Another thing I noticed was that more people were interested in GTK+ than Qt, but more people wanted to develop for KDE than for GNOME... Seemed a little strange.
Well, I noticed it too. Strange, but there can be only one conclusion from this, GTK+ must be ported to KDE!
TeeJay
Avoid the Gates of Hell. Use Linux!
GNOME and GTK+ have really been hyped quite a lot, and I know a lot of Linux users (and even developers) who have used or developed with GTK+ but have never checked out Qt. Also, it's quite possible that a lot of people think of Qt as a part of KDE (since KDE in general is so much more cohesive than GNOME).
I am sure that anyone who is interested in developing object oriented code who's looked at Qt will instantly prefer it over GTK+. It's so much more elegant than GTK+. Aside from the moc hack, which is a little ugly (but it's not something you notice much anyway), Qt and KDE are both quite beautiful libraries.
yes, indeed... what do they do now anyway? I just remember them doing the old turbo c/c++/pascal thing.. (I remember /. posting a link to their site where they gave it away for free now) but I'm wondering what their purpose really is.
hmmmm...
perhaps i should check out their page.
Insert mind here.
What is the most telling question, though, is number 22 - "Currently, the main platform I develop for is...". Windows far out-paces even Linux (which is the closest at only 25% compared to almost 60%). The answers are quickly explained by the fact that they come primarily from Windows developers.
OK. Let's not begin a flame-fest on Winows.
The point here is that many of the Windows developers come from a completely different environment than what spawned Linux. This shouldn't be shocking news. But I do think its being overlooked.
My take on this, which isn't even backed up by an unscientific poll, is that Windows developers are just now looking at Linux. All the market news, conventions, and even Borland's poll has gotten their attention. But they haven't actually looked too deeply into the subject. Linux is unfamiliar waters. Windows developers are not familiar with Linux, its environment, its tools... and most likely, its culture.
What's even more facinating is some of the other questions. Questions involving licensing, source code, development environments, choice of distribution. Many of the answers further show ignorance of the Linux (and Open Source, for that matter) environment... or a very stereotypical "Corporate" view.
There's been talk before on how this attitude would affect the Linux community. Here's an indication that it does, in fact, exist. And it's very likely to interject itself into the Linux world. Is Linux ready? Will Linux change? Or will Linux change the attitudes of developers?
Not strange at all - it just means that people like KDE, but would prefer to develop stuff for it using GTK+.
Personally, I prefer GNOME anyway, but as long as Borland port Delphi to *some* system I'll probably use it. Linux desperately needs something like Delphi.
Although the delphi compiler might not be open source the vcl is.
Wen you buy delphi c/s you get the source code of all delphi components and classes soo you can build your own, or fix those.
regrads
MUHAHA!! now that i have accomplished this feat once, i do hereby solemnly vow to never try again. There is a first and a last time for everything, and this is both.
Well, one reason might be that people wanted to say "I'm equally interested in both of these environments" (and wanted to stress these two, collectively and more or less equally, over others, so they couldn't just leave the question unanswered) -- but there was no such alternative to choose! Hence, go with one on the one question, and the other on the next...
I know that's why *I* answered in something like that fashion.
Christian R. Conrad
MY opinions, not my employer's - Hedengren, Finland.
Christian R. Conrad
mail me at iki.fi ; same user ID as here
see title.
"(they are, after all, Delphi coders)"
What's that supposed to mean, asshole?
Christian R. Conrad
MY opinions, not my employer's - Hedengren, Finland.
Christian R. Conrad
mail me at iki.fi ; same user ID as here
From what I've seen of Delphi, it looks nice, a couple of developers up here swear by it. I guess I'm just not a fan of Object Pascal. I would rather see something like C++ Builder. But then again, that was what the poll was for eh?
Your non-standard usage of terminology confuses the issues -- perhaps because you are confused about them?
... harder than doing it the Delphi way ... fighting the IDE..."
"Object-based" (as opposed to "Object-Oriented") is generally taken to mean "can use -- or fake using -- objects, but cannot create them"; that makes Delphi/Object Pascal OO, and VB Object-based.
"Hybrid" (as opposed to, for want of a better term, "totally OO") means the ability to do stuff the old-fashioned non-OO way *too*, in addition to OOishly. In that sense, the only "totally OO" languages I know of are Smalltalk and apparently Eiffel. Is that what you wanted Delphi to be?
Sorry, can't be done -- if it were, it wouldn't be Delphi any more.
Judging from phrases like:
1) "...go for languages with much better OO support, such as Java..."
2) "...I found it extremely difficult to understand the flow of the program..."
3) "...a bad idea is this whole aspect of building a GUI first, and then attaching bits and pieces of code..."
4) "...avoid this
5) "...I was somewhat disappointed when I actually got to see what it was..."
Make it seem to me as if you haven't really understood it -- "seen", perhaps, but *learned*?!?
Some attempted answers, in short:
1) So what's better with Java's OO support than Delphi's? Maybe you hadn't noticed that the "Beans" component model is pretty much a copy of OP's object model -- maybe you didn't know that Sun actually *bought* it from Borland?
2) Hmmm... Dunno if I can help here... How's this: Think "state", not "flow"? (Yeah, that leads to modal programs -- but hey, I can't say it better!)
3) Well, you shouldn't expect the IDE to give you *all* your code; write the real action in your own procedures (investigate Data Modules and Action Lists), then just call it from your event handlers.
4) Try using it intelligently in stead of "fighting" it. Sorry, but it had to be said...
5) Well, *learn* it in stead of just casting a quick glance.
Christian R. Conrad
MY opinions, not my employer's - Hedengren, Finland.
Christian R. Conrad
mail me at iki.fi ; same user ID as here
"This brings me to my next reason for prefering Qt; it has a very elegant design. Those guys at trolltech know how to execute an OO design in C++. Let the well trained OO designers handle writing elegant OO toolkits (like Qt); the masses (many of whom aren't nearly as well versed in good OO design) use the toolkit and its structure to lead them to a good design."
Yeah, exactly -- and that is what Delphi and C++ Builder are all about: An *extremely* elegant OO toolkit, the VCL, written by the *really* well trained OO designers at Borland. (I mean, what did you think it was -- the base language? C++B's language is just ordinary dime-a-dozen C++, except for one or two added keywords. And yes, dammit, they *were* necessary to enable a RAD IDE, even though they are a kludge. So what -- OP isn't, and C++ was one already anyway...)
And since all the OO elegance you need is in the VCL, you don't *need* that in the widget set it is implemented on top of; a simple API will do -- hey, these tools live on Win32, right?
Another factor was goodwill and "True Open-ness". Sure, QT's license has gotten better lately... But TrollTech *is* still a commercial entity, and as such their product is presumably less "Free" than GTK. But then, in order not to "diss" QT/Troll *too* much -- because it is still definitely the second-best alternative -- I suspect many of us answered like I did, "KDE", on the next question about desktop environments; those are after all a bit less fundamental than the widget set below them.
Now do you see where these percentages are coming from? Seems to me they're perfectly consistent with Delphi coders wanting their tool to come to Linux, *not* being naive or ill-informed but having thought the issues over rather thoroughly, and answering accordingly.
I know that's why *I* answered the way I did, and the overall results are (surprisingly, even to me) consistent with that thinking.
Christian R. Conrad
MY opinions, not my employer's - Hedengren, Finland.
Christian R. Conrad
mail me at iki.fi ; same user ID as here
Another thing I noticed was that more people were interested in GTK+ than Qt, but more people wanted to develop for KDE than for GNOME... Seemed a little strange.
--
Snoochie Bootchie asks:
:-)
"I wonder what makes Delphi such a good RAD tool. I can't believe that the main reason is the use of Object Pascal."
Well, no -- in a word, it's the VCL (OK, that was an acronym, not a word).
"The tool's design and such must contribute significantly. I have nothing against Pascal--it was the first programming language I learned. However, it isn't very popular."
It is with us who *use* it... Does that count for anything, or do you subscribe to the Fly Diet Theory?
"I'd rather use a popular language and become an expert in that language since those skills would be useful anywhere."
OK, have fun in your new career as a Visual Basic programmer!
""How about a Delphi-like took based on Java?"
So what did you think Borland JBuilder is?!? Sheesh...
Christian R. Conrad
MY opinions, not my employer's - Hedengren, Finland.
Christian R. Conrad
mail me at iki.fi ; same user ID as here
Not that I'm saying this is a problem, but I would have preferred seeing a similar represenative survey that was responded to by the community at large.
I mean, development language of choice Object Pascal ? . Note I'm not complaining about OP , in fact I've never used it, I know many people that speak highly of Delphi as a RAD language. I think that its indicative that the majority of resposnse to this survey must have been from existing Delphi / Borland Customers. Mind you they all seem to be slashdot readers as well.
Anyway, I hope they port. seems like there's quite a lot of unhappy windows developers eager to defect.
-- Oh Well
"The real question, though, is why use Delphi when you could use Java? Native java compilers aren't much different from Delphi, other than having the more familiar C++ syntax."
Well, that's one very good reason to go with Delphi in stead of Java -- Java's C++ish syntax sucks, just like it does in C++ itself...
Christian R. Conrad
MY opinions, not my employer's - Hedengren, Finland.
Christian R. Conrad
mail me at iki.fi ; same user ID as here
Forget about vi and emacs. I'm still fond of the old Turbo C, and fte for linux (which is based on the old Turbo editors) is my favourite editor.
It would be nice if Borland-Inprise were to take fte under it's wing and actively develop and make it as universal as vi and emacs.
Question 6: Which Internet technologies are you interested in on Linux?
Their percentages don't add up to 100% (32.6% + 7.7% + 43.1% + 45.6% + 38.1% + 6.2% + 24.1% + 22.1% + 3.5% = 233%)
Question 5 and 7 are the same way.
I don't understand there numbering system...
The answers to questions 9 and 11 are a bit puzzling to me; In 9, GTK+ wins out over Qt by 6.7% (boooo), yet in question 11 developing for KDE wins out over developing for GNOME by about 12.7% (yeah). It makes me wonder how informed some of the respondents were if they didn't know that developing for KDE usually means developing using Qt and developing for GNOME usually means developing using GTK+. And yes, I know, this isn't necessarily the case.
It looks like the Delphi devotees came out in pretty good force; I really didn't expect there to be that many Delphi-heads interested in Linux (or vice versa). I expected C++ Builder to be the Borland tool everyone wanted.
Look at the demographics:
3. The particular development tool I would most like to see for Linux is
12. The Component / Class framework I use should be based on?
22. Currently, the main platform I develop for is...
28. The primary development tool I/we currently use is...
29. All the development tools I/we currently use are...
Conclusion: The vast majority of respondents are Windows users who currently develop with Delphi and want to see it ported to (x86) Linux. It makes sense then that they would use the most Windows-like desktop: KDE.
This sort of thing has cropped up before. And it has always been due to human error.
--
This sort of thing has cropped up before. And it has always been due to human error.
HAL9000
What do you think? Is it just numbers, or is there something to this? Why are so many KDE users (almost half in THIS survey) not interested in developing for Qt?
Several populations are showing up here:
RAD coders: "If I have to think about what toolkt I'm using, you're slowing me down." A good visual GUI builder means never having to care about the widget's code.
Desktop agnostics: People who don't care or even dislike the idea of tight coupling between desktop and apps. I fall into this group myself, unless I wan't some specific feature of the desktop, I'll use whatever widget set it easiest. For me, that means Qt for a C++ project, GTK for C, Tk for perl.
Cross-platform cheapskates: Notice that 60% wanted to develop for both Windows and Linux, with source code compatibility being quite popular. Qt for Windows costs money.
X Newbies: They're still learning the ropes, and don't quite know what's going on yet. Toolkit choice is hardly a big part of Windows programming... They're using KDE or GNOME because their distribution starts it by default.
I'd say this is most of them. Anybody see a group I missed?
The British Magazine PC Plus has had Delphi 1 and 2 on it's cover several times. The Delphi 2 release was the standard version, no RTL source. I believe that the license only allows personal use. Other than that, the distribution is quite usable.
How much of the "pro-Delphi" feel is due to the current lack of popularity of developing in pascal at all on linux, and the relative lack of means to write in it? (Let alone any OO pascal derivatives...)
Why should KDE/gnome and Qt/GTK be regarded as incompatible? I think this poll hasn't helped the polarity there... given you can happily run Qt apps under gnome & gtk stuff under KDE...
~Tim
--
~Tim
--
Rushing on down to the circle of the turn
The survey was available to ANYONE who wanted to take it (it was mentioned on /.). So, I wouldn't necessarily say it was an "Inprise survey". But, the predominant respondents appear to be Windows developers. I would have preferred that a more statistical analysis be done on the data to see what it is that people really want. The repondents were, for the most part, Windows developers. I would have liked to see the data adjust to compenstate for this fact.
What is little known is that Delphi is THE #1 selling RAD development tool for Windows (or so I've read and told). Fact is that Delphi does enable developers to build applications at a rate generally higher than C++. And, it is inherently more robust than VB. The popularity of Linux among Windows developers is growing because Linux is a) stable and b) free c) it's not M$ and finally d) it's a new market to explore for both commercial and open source or GPL's apps.
It's only natural that these same developers want their tools on Linux. And, it's no suprise that these same developers have no clue what toolkits are used by KDE & Gnome. Finally, I am suprised at the number of KDE users since Red Hat makes it so easy to install Gnome.
This would probably explain why people want to develop under linux rather than windows. The Windows desktop gets horribly crowded with millions of toolbars, menus, dialogs etc when developing. Under Linux, I have three windows - Emacs, shell and the application I'm running. That's it. Much cleaner, much nicer.
I'm currently working on some development in IRIX using Motif/Xt, which is so much easier (and infinitely faster) than under Windows. Personally, I find a lot of IDEs counterproductive anyway, and prefer to write and manage my own makefiles etc.
On the subject of widget sets, I've only ever used GTK+, not looked at Qt. AFAICT, however it looks fairly 6 and two 3s between them; what would be cool is an interface layer for Qt->GTK+ or vice versa, but as I don't know much about the architecture of Qt, I can't comment on whether this is possible.
I've rambled, sorry...
-- I reserve the right to be completely wrong --
Interesting that most users are happy to pay for a development environment. Money well spent in my opinion.
The thing with closed-source dev tools, though, it that you can't fix them, and with dev tools you have a very high percentage of users who could fix problems. I'd be happy to pay for tool improvement (for example, I want diff tools that will only highlight the words that change on a line if it's very similar to the line in the file compared to), but I want the resulting tools open. Figuring out a way to do that practically (and in a way the suits will trust) may be a problem. Open source isn't necessary, but available source (and the right to distribute at least small patches) is.
One possibility might be just that. A tool that uses gcc as the compiler, uses makefiles and the autoconf system and just provides an easy-to-use wrapper for the less Linux-savvy could be quite popular.
Clearly the people who took the survey, a self-selected bunch who are interested in Borland stuff, want Delphi for Linux. I would be stunned, then, if this isn't their main product. Or even better would be a base IDE that can work with gcc-based development and has an optional Delphi "plug-in." I know I'd buy it.
It is nice to see that the main interest is people currently doing mainly Windows stuff that want to do Linux stuff. It's a sign that the growth curve for Linux will continue its steep upwards slope.
Ooh, a sarcasm detector. Oh, that's a real useful invention.
Thats probably why the "Traditional IDE" did poorly against the RAD IDE, that could use traditional tools.
Note that command line tools did terribly
-Yarn - Rio Karma: Excellent
Well, for some platforms, closed source compilers outperform gcc/egcs in terms of performance. Obviously this means more work needs to be put into gcc/egcs on these platforms (Alpha comes to mind), but until gcc/egcs can produce code of comparable performance, companies may go with the higher performance, closed source compilers.
Also, from a less idealistic viewpoint, not everyone interested in Linux cares much about whether everything about it is open sourced. A lot of new comers are mainly interested in answering the question "How can I make money off of this?".
There actually is a company that has an ASP implementation for Linux. I think they advertise on Slashdot occasionally.
D
----
They want to develop for KDE because it's the most pervasive Linux windowing environment. Heck, I'm using it right now.
They probably want to develop using GTK because it doesn't require C++. I know that's my position - I'm much more comfortable with regular C.
D
----
You already have C/C++ on Linux so what more do you need??? Obviously there is some benefit to using Delphi. I've played with C++, PowerBuilder, xBase, BASIC, COBOL, and Delphi. I can tell you that Delphi is very robust tool that will give you tremendous productivity. As a consultant once told me... "We have tons of projects going on at any given time. We use many different languages here. Of all these projects, only the ones that are being built using Delphi are on time, under budget, and meeting all of the customer's feature requirements. OP is an EXCELLENT language. Don't discount it.
Just wanted to throw this out. C/C++ lets you do things you ought not to (like reference array element 1000 on a 10-element array... OOPs!) Delphi compiled code runs about the same speed as C code (faster than some slower than others) on NT. There is nothing that you can accomplish with C that you can't accomplish with Pascal. Delphi is a more productive IDE than C/C++. Delphi has state-of-the-art IDE features that the rest of the industry is just beginning to adopt. Now tell me why so many people find that a C++ tool is somehow better than Delphi.
Delphi is a toy. I can't believe the non-creative script kiddies influenced the survey so much. Little good having DELPHI will do for getting Linux accepted in the enterprise.
I was surprised too. When I answered the question I thought back to the good old days (1992) when I was using Borland C++ v2.0
I had some code like this:
float foo;
foo = 4.0
foo += 2.0
After running this code the value of foo was 4.0. I called Borland and asked for help and they blew me off. Then I dumped out the assembler and discovered that the compiler had a bug in it. Then I called Borland back and they admitted their bug.
So, never again Borland! Open source only, thank you very much. I don't give companies that treat me like that a second chance. If you release as open source I don't *have* to trust you.
If tits were wings it'd be flying around.
The original survey had several questions of the "Pick all that apply" type. From a UI standpoint, they used checkboxes, not radio buttons. Thus, the answers reflect what percentage of people are interested in each specific thing, not which thing is the most popuplar. I find that to be a more useful statistic, anyway. Anyone who focuses on just *one* thing and one thing only is asking for trouble. :-)
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
Pascal is as open as C. The choice is between open-source and commercial. As far as that debate goes...If you could get a Porche for $100 or an VW bug for free which one would you choose? If Delphi is 1000 times the product that the open-source community is providing then there will be some that are willing to enjoy the quality even if it does cost a few bucks. I'm a developer for a Fortune 200 company. My time is worth something. If the average developer makes $30 per hour then how long do you think it might take to recoup that $150 to $500 investment? Not too long I suspect. Consider a 20% productivity improvement when using Delphi over open-source tools. If I paid $500 for Delphi and you were using open-source and we each make $30 per hour then you would be loosing $250 per week in productivity after two weeks (the pay-back time for my investment). That adds up in a hurry!
Rarer but not inferior
If you read the questions closely, especially those concerning porting, you see that the majority wants to develop under Linux and port to Windows. This tells us two things:
1) Most developers, even Windows developers, would prefer to develop under Linux.
2) Most developers expect Windows to be more important, in terms of PHB requirements, for the foreseeable future.
This shouldn't surprise anyone. Linux is the ultimate hacker's OS, so of course us hackers like it. It is, on the other hand, scary to all those PHB's, who don't know a bit from a baud.
The cake is a pie
Consider a 20% productivity improvement when using Delphi over open-source tools. If I paid $500 for Delphi and you were using open-source and we each make $30 per hour then you would be loosing $250 per week in productivity after two weeks (the pay-back time for my investment). That adds up in a hurry!
Consider that both your compiler and the other guy's compiler have a serious bug that causes your program to fail mysteriously. You wait a month for Borland to fix your bug and lose $4800, while the guy using the free compiler fixes the bug himself in under a day, losing only $240.
There is something fishy about this survey. I think lots of people who have no clue about development showed up. How else do you explain that they would like to use GTK to develop on KDE...
hmm
I got notice from Watcom - Watcom C++ is no more ...
They stop any development with this product.
How about asking them to release the source code ?
The KDE & Gnome question and cluelessness . . . I use KDE but still chose the GTK+ as a preferred toolkit. I would guess the others who also did so for similar reasons as I. Qt despite improvements to its liscense is still restrictive. GTK+ is not. Whether the different liscenses make a difference to projects I work on (it probably wouldn't) choosing the correct toolkit for the whole platform is **very** important. If Borland came made a mistake in that area and were restrictively bound to a toolkit that was prohibitive to commercial apps they would not sell well. If they only support one GTK+ is the right choice. The optimal would be being able to decide on your own. In RAD development that could be a big hassle for them to include two different sets of code for the same widget sets (although abstracting may be possible it complicates it greatly). Regardless, if they choose one a programmer still has the option to hand code using the other, of course without RAD.
A lot of people seem to be expressing surprise that Object Pascal (OP) got the high votes that it did. And while yes, some of that is due to the fact that Borland's flagship product is Delphi (which is an OP IDE), there is more to it then that.
.EXE for runtime. All an IDE form is, is a component that the IDE automatically feeds an embedded I/O stream to at construction. Very smooth!
:-)
OP, for those that do not know, is Borland's own version of Pascal, extended to support OOP and a number of other things. Unlike MS's Visual Basic, however, it actually comes with a full language grammar spec, and it neatly separates the core language from the RTL (like C and C++ do).
I can speak from experience that OP is *NOT* like ISO ("Standard") Pascal. ISO Pascal has a number of enforced brain damages that make it nearly unusable for real-world development. OP does not.
OP is actually very like C++, once you get past the language grammar layer. Indeed, the back-end optimizer and translator for C++Builder and Delphi are the same. There are a few things OP does not support that C++ does. The biggest are multiple inheritance and templates. Multiple inheritance is useful but relatively uncommon; I don't miss it. The lack of templates really sucks at times, but OP at least gives all objects a root ancestor (TObject), so you can fake some of it. (I sure hope the next OP rev will have templates, though!)
Personally, I find the language design and syntax of OP much, much cleaner and easier to read then C++. C++ sometimes seems like a collection of "const" and the ampersand; the code can look like line noise if you're not very careful. OP looks much neater, even if you're several levels deep in an expression. Part of this is because it uses spelled words instead of symbols for most things, which I agree with. (So does C++, apparently, as the newest spec gives you "or" and "and" operators instead of || and &&).
OP even has some features that C++ does not: Named constructors, properties, and closures, just to name a few. Properties are awesome: They allow you to declare a class member that functions like a variable to class users, but can call class methods when read or assigned. It is much cleaner and more intuitive then Get/Set methods; I miss it whenever I write pure C++ code.
Delphi's exception handling architecture is *much* cleaner then C++. For one, it gives a pre-existing class hierarchy for standard exceptions, and it uses them in a consistent manner. Half the things in C++ throw exceptions on error, half return NULL, and half do some other thing. (Yes, I know that adds up to 1.5 -- I'm making a point!)
Delphi also lets you selectively enable/disable runtime checks for range, bounds, NULL dereference, and the like. This is *great* during development, as it can uncover stupid mistakes and assumptions as they happen. Then you can turn it off for release, and get performance very near that of "bare metal" C/C++.
As for the Visual Component Library (VCL), OP's runtime library and application framework, I love it. It is very elegantly designed, easily extensible, and flawlessly supported by the IDE. It gives the "drag-and-drop" ease-of-use of Visual Basic, but without all the brain damage. I pity the people stuck using MFC -- VCL is *infinitely* better.
The design of the VCL's RAD support is particularly neat. You can create VCL GUIs using the IDE or code or both: They mix seamlessly. (Try *that* with VB!) The IDE can even note changes you make in code and back-update the IDE, and vice-versa. Any VCL component that is persistent (i.e., can be used in the RAD IDE) does so by providing standard methods to read and write its properties to an I/O stream. That stream is saved to a file (.DFM - Delphi Form) for design, and embedded in the
(For those that just refuse to use Pascal, the VCL is also used in C++Builder. The same library, just with C++ bindings.)
Anyway, I hope this little tirade of mine makes the point that OP is not just a VB clone, or another implementation of broken ISO Pascal. I'm not saying C++ is better or worse, just that OP has its advantages too, and is quite usable for real-world development.
</SOAPBOX>
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
Many of you wonder why Delphi came on top instead of C/C++. It was not a surpise for me. If Microsoft would run a similar poll, Visual Basic would be on top. To understand why, you have to look at things from the other side of the fence.
If you're a Windows-user, you have several tools to choose from. You can use Visual Basic, Delphi or C/C++, for example. If you use C and want to write for Linux too, there is GCC, a lot of toolkits and everything you need. You can even use winelib if you want. You'll be productive pretty fast.
But imagine what would happen if you don't do C. What if you have done pascal for 10 years and never touched C? Then there's not any good tools to use. There is Free Pascal, but you can't use it to do X11-programming yet. There is Lazarus but it's not even alpha.
So you're stuck in the Windows-world with all those tens of thousands of lines of VB- or Delphi-code. And you're hating it and praying every day that Linux will get a decent development environment for Pascal or Basic.
Though I find Object Pascal to be quite powerful. I've been a C/C++ programmer for 10 years, and in the late 80's I also wrote a *lot* of Turbo Pascal code.
Pascal seems to have gotten a bad rap, but it's an excellent language for the kind of mindset that RAD requires.
The thing about Delphi (I just started using it last year, and *live* by it for Windows development - it is a far better tool for Win32 development, imho, than VisualC++) is that it's a very, very friendly and productive environment.
Want to get a basic dialog-based app built? Just design it, double click on the GUI elements, add the necessary code to validate each, double click on your "Ok" button, add the final code, and oila - its done.
There is no messing about with Message Apps, no crufty code that doesn't seem to be applicable to anything (as is the case with the stuff generated by the MSVC++ AppWizards), and everything is accessible.
Plus, the Delphi VCL really is an excellent toolkit. When I first discovered it after a long hiatus from writing Pascal code, I experienced the same sort of "gee, programming is fun again!" vibe that I got the first time I studied the BeOS API's... it just makes a lot of sense, and most of the time you can *guess* API function names, sensibly.
Which is not something you can say for MFC or any of the other shitty API's/SDK's you get from Those Folks In Richmond.
So I suppose that the Delphi-cult could best be described as being driven by the fact that Delphi makes Windows programming fun again, something that is sorely missing on the Windows platform courtesy of MS.
; -- the corruption of government starts with its secrets. a truly free people keep no secrets. --
True, but if you know how to program you don't do it that often (ie, use the vector class in STL in your example); besides, Pascal is not type safe (neither C, and I guess neither C++) so you CAN also do things you ought not to in it (I'm not really sure about Delphi, but I guess it's also type unsafe). Besides, you can make wrong programs in any language ... And, I prefer to have a CHOICE, which Pascal usually doesn't give me. If I want to do unsafe things, I can, if I don't I can too.
Delphi compiled code runs about the same speed as C code (faster than some slower than others) on NT
That may be true, but definitely, bounds checked access to arrays is slower than unchecked.
There is nothing that you can accomplish with C that you can't accomplish with Pascal.
True, and you can do everything is Assembly or Basic (or a Turing Machine). The thing is how easy or convenient it is to do it.
I remember programming in TP5.5, and it was harder than in C, then I tried TP7.0 and it was usable, because it was basically C (with begin/end instead of {} :) I mean, it had break/continue, a type for passing arbitrary sized arrays to functions etc.
Delphi is a more productive IDE than C/C++
Definitely, since C and C++ are both LANGUAGES, not IDE's :), so it's true by default (not that it sounds intelligent, mind you :)
Delphi has state-of-the-art IDE features that the rest of the industry is just beginning to adopt.
Probably true, haven't used it since version 1 :)
Now tell me why so many people find that a C++ tool is somehow better than Delphi. In my particular case:
- I prefer C++ to pascal as a language (multiple inheritance, choice of 'unsafe' features, templates, syntax).
- I don't always want to use a visual tool. Most of my programs are console based.
- Delphi is ONLY available for Windows.
This doesn't mean Delphi is bad or anything. I think of it as a good VBIf you had bothered to read the survey results, you would have discovered that 70% of the survey participants are unwilling to spend more than $300 on a complete development environment. It's really ironic that somebody from the Linux community would call 1500 dollars "peanuts", when the whole community is generally known to scoff at any form of non-free software.
Furthermore, please keep in mind that $1500 will only buy you a single license for a single developer. If Microsoft decided to port Office to KDE, how much would they have to spend on Qt license fees alone?
Lastly, why do you presume that everyone who doesn't want to release the source code for his product has to be a commercial developer? This fee can apply to freeware programs as well, if they don't comply with the Open Source mantra.
Choosing to only do Delphi for Linux would be a mistake. Doing both Delphi and C++ Builder would capture major mindshare and reestablish Borland as the tool provider of choice for many people.
Will in Seattle
Based on the poll results, it looks like they'll do a Red Hat distro and any needed changes to get a S.u.S.e. distro. But unlikely they'll do any other distro without support from them. Debian might squeak by for that reason.
Will in Seattle
Borland is now a 10%-owned strategic ally of Microsoft.
Could this survey be an attempt by Microsoft to get exact data on what Linux users and developers are doing and interested in, in order to better subvert them and starve them out?
Maybe this is just paranoia, but...
if I were Microsoft, it's what I would do...
I agree with you about C over C++. Was C++ designed with the intention of being hard to read? Or does Stroustrup just have some of the most counterintuitive notions of any programmer I've encountered?
.)
I like strong typing. I have casts to override when I want to add chars and integers. I detest automatic type conversion and "promotion". Give me no surprises.
Sometimes when I am parsing text I wish I were using C, but the most recent changes in Object Pascal have made that less an issue than it once was.
Almost anything written in VC++ could serve as an example of how not to write. (Let the flames begin.) Why does little or nothing from MS conform to the principles espoused in McConnell's excellent books?
I have no comment on Linux source style, as I have not yet delved into it.
Each time I am bitten by unexpected behaviors (rarely when in Delphi) I crave the good old days of assembly level programming.
I'm not a Basic programmer, and never have been. (See the comment above on automatic type conversion
I'm only welded to Delphi by virtue of production ease and speed. Give me an alternative which doesn't reduce my productivity, and I will use it.
--- Bill
The results you point to aren't all that weird. My take on these responses is that there are a lot of VB developers who'd like to code for Linux, and they'd like to use their current skills. (I remember seeing a tool a while back that did a VB-to-Java translation, allowing VB programmers to shake off the chains and develop cross-platform apps.) BASIC language support in a Borland VIDE might allow a lot of these folks' skills to be portable.
You can run ASP on Linux. ChiliSoft's ChiliASP, I think. And Apache::ASP with modperl.
MS SQL Server needs all the help it can get, so a port to Linux might actually be in MS's favor. (!)
I too find it very encouraging that despite the obvious windows-developer slant, there's a strong interest in releasing free software. There's hope yet.
I think not...(*poof*)
Isn't it proven by now that oss with volunteer peer review is more stable and more secure than bug fixes under proprietary management? If the compiler is not built from open source, to me, the entire oss argument is easily defeated.
Given an increasing awareness of the apparently sudden? (read: unending) desire of U.S. Gov't (when did it start?) to venture into and through each and every family and business computer, and quietly take off anything they want, and given an increasing awareness of security and privacy issues, the day may come when the market demands that every user program be built onsite from an oss compiler. At a minimum, before this part of the survey is accepted as representative of the polled communities, I think it needs to be tested again - maybe via a Slashdot Poll - but tested again.
Whoever can take something off, can put the same thing "back" on again. Open source compilers, with old fashioned chipsets, are here to stay, I hope.
I'd say this is most of them. Anybody see a group I missed? Yes, and I fall into this group. People who believe choosing the correct toolkit will be very important for Linux's success. If Borland does come out with a good to great RAD but uses a toolkit tied to a company many developers will not be able to use the RAD portion (that is hand coding with another toolkit) because of the tks liscense. I use KDE but think it would be dead wrong and bad for Borland to make Qt the only bindings available in their templatized RAD. Being able to select either for RAD would be best, but would be a development nightmare for Borland. GTK+ wins solely on the fact that it is non-restrictive.
1. The Delphi VCL is a class hierarchy built on top of and calling the Win32 api which is in C and of course, also very specific to the Win32 model at the system and process level. It seems that the entire VCL would have to be re-written, and perhaps re-designed to use one of the Linux api's. I have seen no useful discussion of what this would imply - it must be a major project because when Borland did C++ Builder, they preferred to hack the compiler to do Object Pascal rather than re-write the VCL into C++. 2. Can anyone give a pointer to a useful overview of GUI development on linux - comparisons of GTK, Qt, how does a toolkit differ from a class library, etc. I don't need the part about licensing issues, I need a decent overview of the programming issues.
Your post ably demonstrates your ignorance of Delphi. In the Windows environment, there is little you cannot do with Delphi, apart from writing a Windows device driver (and most would rather avoid that horror, anyway.)
I avoided Windows for years because coding for Windows was 90% about building the UI and only 10% about solving the application problem. Delphi makes the UI part a breeze, and frees me to focus on solving the problem.
Also, I can often spot apps which have been developed in Delphi by the clean UI. It's so easy in Delphi that we often spend more time on it than others seem to when using VC++.
Also, C++ Builder provides the same ease of UI design, while supporting all the complexities of C++. It also finds bugs which VC++ does not.
As to corporate acceptance, if you look at the case studies on the Borland site, you will find some fairly impressive names there.
Acquaint yourself with the facts first, then you may have something useful to offer.
--- Bill
So many Linux sympathizers developing on the RAD level with Pascal... who woulda thunk?
I must admit, though, that some form of RAD would be excellent on Linux. I haven't used Pascal since high school (UCSD Pascal on the Apple IIe). I think some C/C++ would fit in better, but I wouldn't complain about Delphi.
'Course, you can't go wrong with Perl. PerlBuilder, by Borland. Yeah.
I suspect that we will see a JBuilder for Linux before Delphi, as I understand that JBuilder 3 is written in Java.
I am sure we will see Delphi for Linux, and I suspect that CBuilder will follow. One open question is whether they will port their compiler or use gcc. Either one would work as well for me, but I suspect that the latter case would draw more interest from existing Linux coders.
As to synchronization of product releases, it would be incredibly difficult. Moreover, CBuilder has all of the C++ complexity and compliance issues to manage, whereas OP is Borland's own lanmguage. Finally, I think each benefits from the other -- they one-up each other over time, so each new release raises the bar for the other product.
Also, for any who don't know, the compiler back end is the same for Delphi and CBuilder. And CBuilder can compile Delphi source (subject to a few restrictions).
Ultimately, the issue of which products migrate to Linux will depend on how many units are sold. If the Linux market snubs tools with pricetags, then they won't get commercial tools. And as should be apparent, Borland will have to go by stages, testing the waters. Anything else would verge on malfeasance.
I hope very much to see Delphi and CBuilder for Linux, and the sooner the better. I also hope to see Delphi for BeOS. Mostly, I hope that this will show Borland the merits of stepping into multi-platform.
For those who may be unaware, Borland has twice before developed tools for other platforms, and both times they lost money on the deal. The first was TP for Mac, and the second was C++ for OS/2. Let's show them that Linux is a good place to be.
--- Bill
6. Which Internet technologies are you interested in on Linux?
Cold Fusion showed up, but iHTML didn't! Damn Marketing!
Consider this: by supporting Object Pascal under CBuilder, they made it possible to use the majority of the already 3rd party components without change. It had less to do with converting the core VCL, I suspect, than with acquiring that very large, and growing, collection of components.
In addition, these components become more truly components when they may be used in two different languages.
--- Bill
Borland still has a large market share. In C++, they have roughly one in three, and the Delphi base is only slightly smaller.
Delphi programmers are loyal and vocal, but often go unnoticed, as they spend more time shipping code that works than yammering about language issues.
Many of us are ready to jump ship from Windows, but are unwilling to suffer the setback of shifting language, OS, and UI tools, all at the same time. We're not all in it for fun, after all, and continuing productivity is a deciding factor when your paycheck hinges on it.
Linux is pretty nice, in spite of some pretty horrible documentation problems (rivaling even those of M$). It isn't God's gift to OSes, in spite of some of the fawning commentary I've read, but it is certainly a worthy alternative to Windows. Frankly, I'd rather have BeOS. But if Borland produces Delphi for Linux, then I'm there, as I know I will have a tool I can trust, and which will spare me from the immediate loss of productivity which comes with switching UI toolkits.
If you think Delphi isn't particularly popular, you've been reading in the wrong places. It's true that Pascal is largely dead apart from Borland's version, but that is largely because other versions were never capable of enough real-world activities. ISO Pascal doesn't even have I/O, after all. Borland has done for Pascal, and now Object Pascal, what a committee eventually did for C++: made it useful, and controlled its form.
Unfortunately, the ANSI C++ standard, overwhelming though it is, fails miserably in some areas, such as in failing to define standards for name mangling, and so portability across tools is still poor.
If C++ were the only tool for programming Linux, I would pass it by. I'm sure there are many others who feel the same. Asking me to bet on Linux for my livelihood, and to abandon a highly productive tool, as well as a UI I understand (even if I don't much like it) is asking me to sacrifice too much.
Linux ain't a cause -- it's a tool. It needs to ba a sharp tool, else why move from the rusty Windows tool which currently pays the rent?
Linux people want to see more people using Linux and programming for Linux, yet also seem to want to force us all to adopt their methods and tools. Oddly enough, that's the M$ philosophy, too. Assimilation. Forget it.
I live with the warts on Windows, and I can live with the warts on Linux, but I will do so only when I have highly productive tools.
Borland, bring on Delphi for Linux! I'm ready.
--- Bill
Uh, and why wouldn't an application of XML be a good tool to define GUIs?
/mill
Instead of trying to make a GUI that caters to both newbies and hackers why not seperate everything from the back end. Define the GUI with a XML application and write the glue code.
The rest of the poll showed an orientation towards windows programmers, none who use C these days.
Many of us may have responded as we did because the licensing of Qt could be a complication. I know that was the deciding factor for me.
As to Gnome vs. KDE, I have used both, and Gnome just ain't ready for prime time yet. KDE has a polished feel, and into the bargain, will be less of a surprise to users already familiar with Windows. That makes my life simpler when I ship a product.
RAD is more than an interest: it's a necessity. Not having RAD means either outrageous prices for application software, or a failing company. The only way a small company can compete today is through RAD.
--- Bill
Actually, I prefer Mandrake and Caldera, so far. I want KDE, not Gnome, and Caldera has the only good installation process I have seen.
Shipping PC based boxes means adapting to change in the environment, and that means using the install tool, not just cloning drives. All of the other installation tools (other than Caldera) just suck. Especially if you select a custom installation.
I don't care whether Linux is open source. I'm not on a crusade. I do care about a viable alternative to Windows, though. And it has to be viable as the foundation for commercial products.
I'm sure that many of us who are programming for Windows share these concerns. This isn't a hobby, or a labor of love. It's a search for a better, cleaner OS that isn't strangling under legacy support and code bloat.
--- Bill
10% isn't control, and I know people at Borland, and they aren't into being M$ windup toys, either.
Look around. M$ has a 10% stake in a lot of companies. Often, as much as anything, it is to help assure the continued existence of at least the appearance of competition. The troubles they have with the DOJ now are as nothing to what would happen if they wiped out Borland and others.
--- Bill
In reading through the survey it became very apparent there are two specific (and somewhat overlapping) groups that responded: 40% are slashdotters (question 33) and at least 76% are already strong Borland/Inprise customers (question 29, Delphi & C++ Builder users). Thus we can't draw any conclusions about whether Delphi is as popular as it would appear on this survey. And there was a non-negligible number of respondents who either weren't answering seriously or did not understand it. (question 20, "what on earth are you talking about?")
I don't know how to account for the number of open source developers (questions 8 and 17). I would count myself among them, but I don't think there were so many. Probably mostly slashdotters again. The most puzzling result was "Object Pascal" in question 12, which did not appear elsewhere on the survey.
The meaningful, relatively unbiased results are: 1) It looks as though everyone wants a RAD solution these days (questions 4 and 2) and 2) most developers who say they want to develop RAD on Linux (most of the respondents) are still using Windows (questions 22-26). This is a clear message to Borland/Inprise to simply continue to improve their Windows tools while making a Linux port.
Development time vs. compiled code speed seems to be interesting. I tend to believe that time spent programming is more expensive than computing time nowadays. Delphi generates fast code, however!
Hmm, why there was so little about Visual Basic? Oh yes, that was Inprise's survey. Or perhaps VB is in fact a non-hyped dominant language?
God did not appoint us to suffer wrath but to receive salvation through our Lord Jesus Christ --1Thes5:9
Don't forget, folks:
That being said, I think that there is some useful conclusions to be drawn. For example, I seriously doubt that the Linux distro distribution is in any way correlated with this sample's non-representative origin -- although of course it is possible that Borland users tend to prefer RedHat more than the general Linux population.
--
--
Victor Danilchenko
I might consider it if the source was held in escrow, but the conditions for source code release to me would have to be very favourable. It's worth noting that I personally don't have the compiler expertise to be able to fix most problems that are likely to occur anyway. However, if I have the source, I can at least pay someone else that does have the expertise to do it for me.
BTW, it's nice to see that a company like Borland/Inprise still has a sense of humour (re: question 35).
"The invisible and the non-existent look very much alike." -- Delos B. McKown
i downloaded a program and installed it. when i typed ./configure i dont remember checking for gdelphi .... ok those results seemed real weird. i must have been out of town when the survey was posted. i didn't realize delphi dev was that big of a deal. i also agree with the gtk/qt gnome/kde conundrum. i guess alot of people in europe use delphi?
-- john
Delphi tends to be on top. I would've thought that C++ builder would've been more popular with Linux users/coders. Cuz you get C++ code out, not the odd delphi stuff. But even more interesting is some of the other numbers...
~500 people want VB on Linux
~4250 people want ASP for Linux
~3820 people want Microsoft SQL Server on Linux
Ahh the telling point
Which toolkits will you use?
~6250 Don't know what these toolkits are.
~14000 primary develop Windows
9332 deleop primary in Delphi
Good to see that
~5700 will release software free
But it is interesting to see that this list is mainly from windows developers not from Linux developers which kinda explains the other results.
-cpd
Very clever. :)
By filtering everyone out who visits slashdot.org
they can get truthful picture of the situation
instead of heavily biased pro-linux view..
At least that is what I would do
... who has worked in many languages and OSes -- and come out of it with wierd preferences?
/. crowd throws at him.] Yeah, yeah, dis away... I don't care.
Right now I'm working on a Delphi project. To me Delphi is okay... but just okay. Perhaps it's because I've seen everything I'm doing with Delphi done several different ways already in other languages/products. To be honest, for Windows development I prefer Visual Basic. [The AC deftly dodges the amazingly huge number of daggers the
For *nix, I prefer C. NOT C++. I've worked with both, and I hate all the crap that C++ makes me go through (apparently the language was built with the assumption that any other programmers on the project are idiots who I dare not trust with my precious code, my precious my precious). C lets me do whatever I see fit to do... am I the only C programmer in the world who can figure pointers out? It ain't that big a deal, and sometimes I want to add characters to integers, damn it! (g)
I feel like I'm at both ends of the bell curve: I get laughed at for wanting to use VB -- for which I'm assumed to not be a programmer at all; and laughed at for wanting to use C -- for which I'm assumed to either be unable to figure out OOP or that I write unmaintainable code.
Everyone who has a favorite language/product to which they've welded their career seems to laugh at anyone who does something else... perhaps it's a general human thing. The laughter seems so nervous at times, though....
On the other hand, I don't see any commercial outfits flocking to GTK. A development kit isn't just software -- it's a package that includes documentation and support. QT has this. GTK doesn't.
Funny how I am totally opposite in preference
:) I guess :)
towards using spelled keywords as opposed to
symbols.
I find it much speedier to read code, especially
someone else's, in C as opposed to say more wordy
languages like Pascal and Visual Basic. Syntax
highlighting in the IDE helps a lot, but when
there are words all over the place, I find myself
delayed by having to read them, like reading a
story where you don't want to accidentally skip
over important words.
It's mostly a minor issue, but I find myself able
to get an idea of a block of C code faster in
situations where a lot of conditions are being
evaluated. Probably because I don't have to go
back and make sure that I correctly read the
Not's and And's and didn't mistaken them for
variables or something. There's no confusing an
&& symbol from an || symbol (at least for me).
Even having said all this, too much symbols is
also confusing and also slows me down
I have too many problems
On another note, I think the "Then" keyword after
an "If" condition is very annoying and waste of
time to have to type out. Grrr.
Ok, that's enough from me.
Not necessarily. I could see Troll Tech licensing QT to Borland, and collecting some royalties. If Troll had reason to believe that such a toolkit would be a big seller, it would make sense for them to sell it at a lower price. This would be a win-win-win: Troll rake in money in on the kit (which would generate a lot of commercial QT sales ) , Borland can use a stable kit that is carefully documented and maintained by professionals, and the users can develop with the best available toolkit.
Borland need something that's good, not something that's cheap ( note that the respondents said that they were willing to pay ). This in itself rules out GTK.
Really. It looks like another lousy attempt to advert attention to Inprise.
1. Every decent IDE we have under Linux today is configurable to work with almost any existing compiler. Do you need to throw $ 1,500 to the wind to tie yourself to a proprietary thing again? This amount makes a cool amplifier to me.
2. To mean something for the whole Linux community, one needs to span his wings above everyone. Or almost that wide. The vast majority of poll participants are current Windoze developers and they want this or that (language, widget set, wm support etc). How about others? Is Inprise about to conquer Linux or they are just 'porting' developers?
3. Forget about VCL. This Linux, it comes from Unix. Everything you need is already at your finfertips. Probably hiding CreateProcess() behind class method saves typing, but man, execve() is as simple as baby's ass. Keep it simple.
4. Beware. There's a greater part of Unix developers and they have no need to port anything. They have zillions of lines of Hi-End code that works on any hardware and with any widget set. If you only stuck with proprietary tool- they are light years ahead and you loose before you start.
I am with Borland since TP 4, don't get me wrong. But that was Borland. Nowadays Inprise has nothing common with that small, brave and innovative company of our early days. It started for me with 4 and is should end up with the same number. Really, Delphi 4 is a bloatware.
In general, fashions come and go. People can dream away but it will take very long time for Inprise to come out with anything decent for Linux. If they ever manage to. Selling themselves to M$ wasn't a Good Idea (tm). "But look, 20 developers can't feed 150 'kind of' developers and 200 managers, ya know...."
KuroiNeko
First off, what was really interesting is that more people were interested in developing for GTK+ than for QT, but at the same time, more people wanted to develop for KDE than for GNOME. Maybe KDE should adopt GTK+ as their library. :)
:) And Rapid Application Development really seemed to be an interest in most of those polled.
I was also surprised at the popularity of Delphi. But I have to admit that when it comes to RAD, Delphi is RADer than any other C++ tool.
I like Delphi, least what little I've used it. One of the few programming languages I can get a grip on. Borland C/C++ would be damn nice to.
Also, the QT lib is available on both Linux and Windows. Don't know how good the windows port is though.
Linux looks like something conceived by system administrators for system administrators. Therefore, it is increasingly succesful on the server-side.
On the client side,however, when a linux person says *apps*, he means gcc, emacs, sed, awk, perl. When a user says *apps*, he means the things he uses to manage sales orders, time sheets, bills of material, and other stuff that he knocks in to his RAD-tool developed, mostly client/server database-oriented programs, quite often custom-made or adapted.
When I look at the linux community, I can see a quite a lot of idealistic people, who, unfortunately don't have a clue of why these users don't want linux, not even for free. By far and by large, they prefer to shell out cash, and massively, to Micros~1.
Linux, is nothing more than a total waste of time for these users. Time being money, the cost of using linux is, therefore, much higher than all license fees for buggy M$ gear.
If you don't believe me, ask any business, small or big, to manage sales invoices and VAT returns with emacs; or anything else in the latest Red Hat distribution. I guarantee you that they'll even pay you to throw the CD out of the window, and fast.
Seriously, this was an out-of-court settlement.
So, MSFT gave them the money owed, in return for non-voting preferred stock. So if Borland does well, MSFT gets dividends. And Borland needed the cash, so let's not complain.
Will in Seattle
I've used Delphi and C++Builder (several versions of each) for a few years, so here is my perspective on this.
From the looks of this survey, it definitely looks like Borland will consider Delphi for Linux. I like Delphi (from an OO standpoint I think it is fairly strong) but I don't like ObjectPascal.
On the other hand, C++Builder is a great product IMO, but the biggest drawback is that Borland treats it as a second-class citizen. They upgrade Delphi about every 14 months or so, but C++Builder lags way behind....if C++Builder 3 came out only a few months before Delphi 4. Why isnt the development synchronized? Also, all the libraries and stuff are geared towards Delphi; C++Builder is an afterthought. I ran into this problem when I was unable to use one VCL (the Delphi/C++Builder components) because of problems in the C++Builder Winsock code. The component worked fine in Delphi, but wouldn't work in C++Builder. Problems like that made me reconsider using C++Builder over Delphi, but the simple fact is that I like C++ a lot more.
Let's hope that Borland does the intelligent thing and develops Delphi and C++Builder for Linux.
People keep saying that this must be lots of Windows developers, but question 13 makes me wonder about that. Linux application user interfaces should be... Native GUI got >76%!
That's impressive. Even if the people taking the poll were from the Windows school of programming, at least they haven't been brainwashed into thinking that HTML/XML (8.6%) is a good way to write applications. (There might be worse methods for doing application user interface, but I'd have to work really hard to come up with one.)
-Brian
Interesting that most users are happy to pay for a development environment. Money well spent in my opinion.
:)
There's also going to be a big bunfight on the relative merits of Delphi v C++Builder. Having written applications, (for not much more than beer money) in both, let me chip in my 2p's worth. C++ is by far the vastly superior language, make no mistake. Used properly C++ is extremely maintenance friendly and syntactically sweet to the point where you nearly call a dentist. But with the wrong libraries (particularly MFC) it's just a complete pain. VCL is one of those wrong libraries. It is just *so* designed around object pascal that using it in C++ with the yukky CBuilder hack arounds almost defeats the purpose entirely.
I have no doubt that in all but the most extreme cases the object Pascal optimiser will make it good-as-dammit as fast as C++. VCL is a beautifully designed class library and the questions Borland were asking (which back end would you like to use, given that BDE is awful) give the impression that they're quite willing to put the work in to ensure it gets ported to Linux as well as it ought to be.
The Windows 'community' shunned Delphi ages ago, when the PHB's were FUD'ed into using Visual Basic and anyone (read most) who wanted to pay their bills went with them. Great loss, believe you me.
Get yourselves a Win32 box from somewhere and a copy of Delphi 2 off the front of a Magazine. Have a go. It is good, you will enjoy it, and when you discover you can put not insignificant apps together in a week (with sufficient focus) then the need for LinuxDelphi will become aparrent.
Dave
I write a blog now, you should be afraid.
I see even classic Slashdot is now pretty much unusable on dial up anymore.
I found it interesting that when asked what desktop they were using, 50% replied KDE, while only 27% replied GNOME. Being a GNOME user myself, I was kind of surprised at how many are using KDE (maybe I'll check it out).
But the real interesting thing was the toolkit question: which toolkit are you interested in developing in? Only 18% replied Qt, while 25% replied GTK. While a good 26% said they didn't know what all this toolkit mess was about.
So it appears (note, this is conjecture) that possibly 30% of the people who use KDE don't know what toolkit KDE is based on. Possibly--I don't want to draw too many conclusions from the numbers.
Why would this be? Do KDE users simply have it easier--i.e. KDE is such a integrated package that knowledge of the underpinnings are not required, etc.? Being a GNOME user, I am very aware of GTK. One, because of the rapid GTK development in the GNOME 0.9 stages, which broke entire versions of GNOME. Two, because of all the initial "political" hub-bub about GNOME being true opne source because of the GPL v. TrollTech/Qt.
What do you think? Is it just numbers, or is there something to this? Why are so many KDE users (almost half in THIS survey) not interested in developing for Qt?
"Doubt your doubts and believe your beliefs."
"Doubt your doubts and believe your beliefs." -- Switchfoot, Ode to Chin
Somehow the results dont really fit together, do they ? For example take the answers to 9. and 11.. People want to develop for KDE but use Gtk+. Somehow this doesnt really go together.
Also with kdevelop progressing that fast and with the current integration of standard tools, who needs Borland anyway ?
There is no need for them to be. So they used KDE and probably heard more about GTK.
Actually, one of Delphi's biggest problems lies in exceptions... Unfortunately, it doesn't provide guaranteed construction/destruction, which is why it has that horrible "finally" kludge. In C++, you don't have to scatter such hacks all through your program. The lack of genericity (templates) is, indeed the other big problem. The real question, though, is why use Delphi when you could use Java? Native java compilers aren't much different from Delphi, other than having the more familiar C++ syntax.
Think about the people who use Delphi and want to move to Linux... They're that 95% (or is it 90?) that ESR talks about who develop internal software. They don't care if it costs $1500, since they don't pay for it themselves. They want a simple way to develop software quickly, and that means RAD tools and big libraries. They don't care if it works on someone else's system, since they're not developing for someone else's system.
I'm in a hurry, so I haven't read all the comments... feel free to moderate down if it's redundant.
Now, these have got to be the most clueless survey results I've ever seen! From the looks of the answers, at least 75% of the responding persons had no idea what they were talking about.
Let's take a look at the "developer" you would get by combining the highest scores:
"What I want is to write Free Software, but I don't think it's important for people to be able to tweak my code. It will, however, most definately be GPL'd. Now, I only use KDE, and have never heard of GTK+, but that's the library I want to develop with. For KDE. Not that I know what GTK+ is. I will be writing these Free, binary only, GTK+ based, KDE apps (definately GUI) as server tools, for a server running Oracle. But I want no profit from the customers I have (that are demanding my stuff runs on linux) even though they can afford Oracle."
All I want to know is, what kind of glue do you have to sniff to get there?
Ahem. Sorry:
"Not mostly server side desktop tools, rather mostly desktop apps. That use Oracle."
Hehe. Not much better ;)
(I hope I didn't mess up before by just hitting 'enter' after typing the subject... I might have submitted an empty comment ;-)
;-)
;-)
:-)
A couple of things stood out to me and one was that most developers don't seem to know what the toolkits for GUI development are.
From this I can draw the conclusion that they are mostly windows-developers, and that shouldn't be much of a surprise since most of Borland's products are windows only at the moment.
However, since KDE is pretty much a default environment in most distro's I'm not surprised to see many developers say that they use KDE and want to develop for it.
But being windows developers, they don't yet realize that this means Qt development: They have no clue about the toolkits underlying Unix GUI development.
("What, that's just a class-library, isn't it?")
Another very interesting point was, imho, that most developers wanted: Easy porting of Linux -> Windows, rather than win->linux.
In other words: They want to develop primarily on Linux, and then port that to windows as a secondary development platform.
Could it be that application development for Unix/Linux is really that much easier than windows app. development?
Is windoze being replaced as a primary, preferred development environment?
Anyway, I hope that Borland comes out with something nice and sweet!
--Tim