Opinions on KYLIX 3 (Delphi 4 Linux)

[windozbloz]

Bye Bye Billy Bob...

I'm back with one more question, then I'll chill. I have scoured the news
and net for info about Borlands KYLIX 3 and have found little technical
info about it. Their screen shots are very impressive, similar to Visual
Basic. I have sent several emails to Borlands Sales and Pre-Sales
departments. Pre-Sales bounces and Sales won't answer. I'm sitting here
with money in hand ready to buy but not from a company that won't give me
the time of day.

Does anyone of you have experiance with KYLIX 3 and do you think I should
consider buying it? Thank You, I'll go oil my keyboard now.

Doug

 

[Jeff Epler]

I honestly don't know why anyone would spend money for a development
environment, no matter how fancy. I don't know why anyone would develop
software in a language that doesn't have at least one open
implementation.

It's a great way to get screwed when Borland goes under or decides
they only want to sell a new, incompatible product. What do you do with
your existing product when that happens? Re-train on a new platform,
and re-write from scratch?

Just say no to proprietary software.

Jeff

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQFC29uaJd01MZaTXX0RAnjcAJ4/2ZG2kfV2oVjhLPhBr06t7kVr+QCdEPYU
nbTIrMSqZfAkB/xQ1uUg+eA=
=Nisl
-----END PGP SIGNATURE-----

 

[windozbloz]

I honestly don't know why anyone would spend money for a development
environment, no matter how fancy. I don't knowdefinitelye would develop
software in a language that doesn't have at least one open
implementation.

It's a great way to get screwed when Borland goes under or decides
they only want to sell a new, incompatible product. What do you do with
your existing product when that happens? Re-train on a new platform,
and re-write from scratch?

Just say no to proprietary software.

Jeff

Thanks Jeff,
Point taken! I had given that considerable thought. Your words 'retrain on
a new platform' struck a loud cord with me, thank you, I will definitely
reconsider. I don't ever want to find myself locked behind 'the GATES of
hell' again!

Doug

 

[Thomas Bartkus]

Bye Bye Billy Bob...

I'm back with one more question, then I'll chill. I have scoured the news
and net for info about Borlands KYLIX 3 and have found little technical
info about it. Their screen shots are very impressive, similar to Visual
Basic. I have sent several emails to Borlands Sales and Pre-Sales
departments. Pre-Sales bounces and Sales won't answer. I'm sitting here
with money in hand ready to buy but not from a company that won't give me
the time of day.

Does anyone of you have experiance with KYLIX 3 and do you think I should
consider buying it? Thank You, I'll go oil my keyboard now.

Good question! Wither Borland?

My impression (second hand - based on no direct experience with Kylix!) is
that Borlands wonderful Delphi product ported to Linux has been a
dissapointment.

* * * Someone with real experience on Kylix - please jump in here!

Calling Delphi "similar to Visual Basic" is hurtful because I believe that
VB is the product of looting and pillaging the talent that came out of
Borland. I'm guessing that Microsoft has successfully targeted this
perceived competitor with destruction.

If Kylix were of the quality of Delphi, it would be a killer Linux app.
Thomas Bartkus

 

[Thomas Bartkus]

It's called (ROI) Return On Investment. If you can get that to a positive
number, then spending money on the tool is a no brainer. Given that one
caveat, a sane person will just do it!

If I can create something that earns $10,000 or more in profit, then the
difference between a tool that costs $500 and one that is free amounts to
small beer. IOW - it hardly matters! The only decision one needs be
concerned with is which tool is most effective.

Your existing products are not affected. The compiler tools you have will
work just as well as the day you bought it. Your ownership of the product
does not get revoked just because Borland dissapears.

What do you do when an open source project you were using gets abandoned?
Hard to see much difference here.
Thomas Bartkus

 

[Caleb Hattingh]

Hi Doug

Not only was Kylix a letdown, there is talk also of it being
discontinued. To be fair though, it is easy to see the difficulty for
Borland to deploy a Linux IDE of the same quality as Delphi when so much
in different Linux distributions is variable, the widget set being a prime
example: CLX just never caught on, amongst the myriad of more mature and
more open toolsets.

I am assuming that you have experience with pascal, or ObjectPascal as the
Delphi manuals call it (if not, why Kylix?). If so, may I suggest you
look into
a) fpc (Free Pascal Compiler, http://www.freepascal.org)
b) Lazarus (An fpc IDE, aims to be an open-source Delphi clone, supports
multiple widget sets [Win32 native on Windows!], can't remember the URL)

Though python is probably my favourite language for general purpose
hacking, there is a lot to be said for a native compiled language
(sometimes speed _is_ an issue), and in particular there is a lot to be
said for fpc:
- Something like 99% Turbo Pascal compliant
- Very nearly Delphi compliant (object-pascal)
- (IMHO) Much cleaner language than C, still natively compiled
- Supports operator overloading & inlining
- Can be used to develop python extensions
- Supports nearly effortless cross-compiling

There are some problems with Lazarus at the moment, here is the biggie:
It creates binaries of around 5MB for a minimal app, and this is because,
at the moment it seems like it compiles the component library into the
executable. This is somewhat of a problem for me but I expect this will
change within a release or two.

Lazarus is _very_ much like Delphi, and works on Windows, Linux, and
possibly several other platforms.

cya
Caleb

 

[Luis M. Gonzalez]

I'd suggest you take another path:
Since we're in a python mailing list, I assume you like python, so
instead of Kylix (or Object Pascal), you could try Boo
(boo.codehaus.org) .
Boo is not python, but it is very similar.
It could be described as a statically typed version of python for the
..NET framework (it also works in Linux with Mono and in Mac Osx).
It has borrowed from python as much syntax as possible and uses a handy
type inference system, which allows you to code without having to
declare types everywhere, making it very similar to a dynamic language
while being static.
As I said, "it is not python" (Peter Hansen, please don't jump to my
jugular...), but it's been inspired and takes a lot from it. In words
of Guido Van Rossum, it is 95% python.
You get all the niceties you got used to enjoy in python, such as list
comprehensions, generator expressions, slicing, dictionaries, lists,
etc, etc.. and you get all the speed and performance of a static
language for free.
As for IDEs, you can use SharpDevelop in Windows (free) or MonoDevelop
in Linux (free).

And if you are patient, soon you'll be able o use regular python in
..NET and Mono, and I'm sure that there will be support for SharpDevelop
and MonoDevelop for it as well.
At the present time, python for .NET (code named Ironpython) is still
at pre-alpha state (version 0.7.6).

 

[Mike Meyer]

What do you do when an open source project you were using gets abandoned?

Hard to see much difference here.

Doing support for object-only distributions is *much* harder than
doing it for source distributions.

I have a habit of picking products based on technical superiority, not
popularity. As a result, I have a nice collection of orphans. That's
because technical quality has little or nothing to do with
profitability.

On the other hand, since starting to use open source projects, I've
never had one I depend on fail. I've had some I contributed to fail,
but that's a different thing.

I suspect that technical quality in open source projects contributes
to their attracting people to support them. This makes them ever so
much more attractive than proprietary solutions, where technical
quality seems to be irrelevant to longevity.

<mike

 

[Thomas Bartkus]

cvs import -m "sources for orphaned project" <myprojectname>


Doing support for object-only distributions is *much* harder than doing
it for source distributions.

I have a habit of picking products based on technical superiority, not
popularity. As a result, I have a nice collection of orphans. That's
because technical quality has little or nothing to do with
profitability.

On the other hand, since starting to use open source projects, I've
never had one I depend on fail. I've had some I contributed to fail, but
that's a different thing.

I didn't suggest that orphaned open source projects were a problem. I
simply point out that they are no more, nor less, of a problem than an
orphaned (and paid for!) commercial product.

I suspect that technical quality in open source projects contributes to
their attracting people to support them.

Perhaps. And there is no way to support a commercial product other than
by becoming an employee.

This makes them ever so
much more attractive than proprietary solutions, where technical quality
seems to be irrelevant to longevity.

This last statement sounds too much like a canard. It is difficult to deny
that commercial products either put some significant value on the table or
go bust. Although people can be, and sometimes are, swindled few can
afford to simply throw their money away. IOW - technical quality is
*never* irrelivant to longevity. And one must also consider that
technical merit, by itself, is rarely sufficient. The open source world
is awash with much that is high on technical merit but commercially
unviable. There is much out there that one would gladly pay good $ for if
only for lack of that last (but most difficult!) 5% effort that would
bring many of these projects to fruition.

Which brings me back to the point that the difference between free and
$500 (or $1000!) amounts to virtually *nothing* when evaluating a tool.

You use what is most effective to get the job done.
Thomas Bartkus

 

[Luis M. Gonzalez]

As I said, "it is not python" (Peter Hansen, please don't jump to my
jugular...)

Hey Pete, I was joking on that one!
Here's the missing smiley ;-)

 

[Mike Meyer]

I didn't suggest that orphaned open source projects were a problem. I
simply point out that they are no more, nor less, of a problem than an
orphaned (and paid for!) commercial product.

You missed my answer to this. Ok, it was oblique, so it's probably my
fault. With an orphaned open source project, you always have the
option of taking on the support role yourself - or paying someone else
to do so. That's not generally possible with closed source products.

Perhaps. And there is no way to support a commercial product other than
by becoming an employee.

Not true. Not all commercial products are closed source - though that
tends to be the norm these days. And you can contribute code to closed
source products. For instance, I did the original Python wrappers for
Perforce's binary library. That's certainly supporting the product.

This last statement sounds too much like a canard. It is difficult to deny
that commercial products either put some significant value on the table or
go bust. Although people can be, and sometimes are, swindled few can
afford to simply throw their money away. IOW - technical quality is
*never* irrelivant to longevity. And one must also consider that
technical merit, by itself, is rarely sufficient. The open source world
is awash with much that is high on technical merit but commercially
unviable. There is much out there that one would gladly pay good $ for if
only for lack of that last (but most difficult!) 5% effort that would
bring many of these projects to fruition.

I almost certainly overstated the case - but I've been burned a lot by
choosing technical quality over profitability or popularity. The
reality seems to be that once you reach the level of "good enough",
technical quality stops mattering, and marketing forces come into
play. My essay on the subject at <URL:
http://www.mired.org/home/mwm/good-enough.html > has more information.

Which brings me back to the point that the difference between free and
$500 (or $1000!) amounts to virtually *nothing* when evaluating a tool.

Depends on what you're evaluating it for. In the context of the
discussion - choosing tools for commercial software development - it's
almost certainly true. If it's not, you're not charging enough for
your time. But that's hardly the only time that one evaluates tools.

<mike

 

[David Trudgett]

Good question! Wither Borland?

My impression (second hand - based on no direct experience with
Kylix!) is that Borlands wonderful Delphi product ported to Linux
has been a dissapointment.

* * * Someone with real experience on Kylix - please jump in here!

It has been two or three years since I gave Kylix a try, so my memory
is a bit vague on the specifics. I was working in a Delphi shop and
wanted to port (at least some of) our apps to Linux using Kylix (I
think it was version 3). I think I ported one and a half apps and more
or less gave up or put it on the back burner. My impression was that
Kylix still wasn't ready for serious development work.

The type of application I was working on (porting) involved
client/server database access, and TCP communications with other
applications. It never really worked correctly (I forget what the
problems were just now), but probably could have been made to work
correctly. The point was, however, that porting (a relatively simple)
Delphi app to Kylix shouldn't have been that hard.

Calling Delphi "similar to Visual Basic" is hurtful because I
believe that VB is the product of looting and pillaging the talent
that came out of Borland. I'm guessing that Microsoft has
successfully targeted this perceived competitor with destruction.

If Kylix were of the quality of Delphi, it would be a killer Linux app.

Possibly. Unfortunately, I don't believe that the whole GUI building
approach of Delphi/Kylix (or other similar tools) is much chop. It
encourages one, for instance, to just place elements on the screen in
fixed positions that make no allowance for differing fonts, screen
resolutions, etc. Java (my experience is with JBuilder) is much better
in this regard, although the different paradigm takes some getting
used to. However, all GUI builders with which I'm familiar (not many)
seem to have very real limitations when it comes to designing very
complex interfaces. Kenny Tilton's Cells project (ask on
comp.lang.lisp) has set me to thinking along these lines. In the past,
I never gave it much consideration.

Programmers who like Pascal should look at Ada as a better
alternative. If I wanted to program in a Pascal-like language on
Linux, Ada (the GNU Gnat compiler, integrated with GCC) is the one
that I would use. Ada, you could say, is like Pascal on
steroids. Caveat: I've read Ada books, but haven't programmed in it,
and my main concern is that its ultra strong typing might get in my
way -- or alternatively, force greater rigour, as the Ada folks might
say ;-).

These days, for hacking about, I prefer Common Lisp. It's faster
(sometimes approaching the speed of compiled C/Pascal) and much more
powerful than Python, but doesn't have the same library support
(smaller community), and application bundling and delivery *can* be a
potential problem, depending on various factors (such as whether you
want to license a commercial Common Lisp). Also, similar to Python,
there is no standard GUI framework defined for Common Lisp, so
choosing from the GUI frameworks available can be a challenge (I've
only programmed a simple GUI app using the great little Ltk library by
Peter Herth, which talks to Tk over a socket).

My advice would be to steer clear of Kylix and choose one of the other
environments suggested to you. If you really like Pascal, fpc may be a
possibility as someone mentioned. I haven't looked into it any time in
the last couple of years, though, so I don't know its status. I really
would suggest a serious look at Ada, though, if you want to develop
fast, industrial strength applications, or take advantage of built-in
concurrency support and lots of other goodies.

David



--

David Trudgett
http://www.zeta.org.au/~wpower/

I was in the pub last night, and a guy asked me for a light for his
cigarette. I suddenly realised that there was a demand here and money
to be made, and so I agreed to light his cigarette for 10 pence, but I
didn't actually give him a light, I sold him a licence to burn his
cigarette. My fire-licence restricted him from giving the light to
anybody else, after all, that fire was my property. He was drunk, and
dismissed me as a loony, but accepted my fire (and by implication the
licence which governed its use) anyway. Of course in a matter of
minutes I noticed a friend of his asking him for a light and to my
outrage he gave his cigarette to his friend and pirated my fire! I was
furious, I started to make my way over to that side of the bar but to
my added horror his friend then started to light other people's
cigarettes left, right, and centre! Before long that whole side of the
bar was enjoying MY fire without paying me anything. Enraged I went
from person to person grabbing their cigarettes from their hands,
throwing them to the ground, and stamping on them.

Strangely the door staff exhibited no respect for my property rights
as they threw me out the door.

-- Ian Clarke

 

[Thomas Bartkus]

On Wed, 20 Jul 2005 18:37:48 +1000, David Trudgett wrote:

My advice would be to steer clear of Kylix and choose one of the other
environments suggested to you. If you really like Pascal, fpc may be a
possibility as someone mentioned. ...

<snip>

Well - I really like Python!

But - climbing back on the soapbox - far more important than any
linguistic quibble is a robust GUI interface that might *unburden* any
language from the tedium of creating a powerful and effective user
interface.

I would go so far as to say that, while this is the only thing an atrocity
like VB has going for it, it *trumps* most everything. This was the great
hope of the Delphi<->Kylix thing. The Windows implemntation (Delphi)
was/is a magnificent productivity tool. And if Kylix has failed then we
are still scratching about with TKinter, Glade, wxGlade, EasyGUI, or
{whatever} just to to bring up the rear on Linux.

Python is fun, but how do we get it to desktop primetime ?
Thomas Bartkus

 

[Peter Maas]

I honestly don't know why anyone would spend money for a development
environment, no matter how fancy. I don't know why anyone would develop
software in a language that doesn't have at least one open
implementation.

FreePascal is OSS. I recently developed a mixed Delphi/FreePascal
application. FreePascal doesn't have a GUI Builder like Delphi but
is very complete and mature.

It's a great way to get screwed when Borland goes under or decides
they only want to sell a new, incompatible product. What do you do with
your existing product when that happens? Re-train on a new platform,
> and re-write from scratch?

Port it to FreePascal

 

[Vladimir Konrad]

Does anyone of you have experiance with KYLIX 3 and do you think I should
consider buying it? Thank You, I'll go oil my keyboard now.

http://www.lazarus.freepascal.org/

Sorry if the above came out somewhere else in the thread, did not read all.

Vlad

 

[Ruben Baumann]

Bye Bye Billy Bob...

I'm back with one more question, then I'll chill. I have scoured the news
and net for info about Borlands KYLIX 3 and have found little technical
info about it. Their screen shots are very impressive, similar to Visual
Basic. I have sent several emails to Borlands Sales and Pre-Sales
departments. Pre-Sales bounces and Sales won't answer. I'm sitting here
with money in hand ready to buy but not from a company that won't give me
the time of day.

Does anyone of you have experiance with KYLIX 3 and do you think I should
consider buying it? Thank You, I'll go oil my keyboard now.

Doug

KYLIX is currently not being developed by Borland. I'm not even sure
they're still supporting it OR selling it. There is a guy that is currently
providing patches for the product(third party), but KYLIX 3 is the last
version so far.

I suspect that Borland is going to eventually drop KYLIX in favor of .NET
and it's c# product. There are strong indications that the MONO project
will run vb.net, c#.net, etc, etc, etc, so there's no need for a delphi
(pascal) compiler under Linux if true.

HTH
Ruben