Board index » kylix » RE: Kylix now officialy dead?

RE: Kylix now officialy dead?


2004-11-22 03:37:02 PM
kylix1
Quote
I'm about to start writing a remote de{*word*81} for CrossKylix. First beta
tests hopefully will start at the end of November.

Great !
I really wonder how that can be done ! Please keep us posted.
Will it be usable from the Delphi IDE like when remotely debugging on a
Windows target, or will it use a different frontend ?
-Michael
 
 

Re:RE: Kylix now officialy dead?

Michael Schnell wrote:
Quote
Agreed !!! ANDREAS PLEASE !!!!! :)
Me? Or do you mean Simon, who wants to write a remote de{*word*81}.
--
Regards,
Andreas Hausladen
 

Re:RE: Kylix now officialy dead?

On 2004-11-22, Michael Schnell < XXXX@XXXXX.COM >wrote:
Quote
>Feel free to google for all glibc and kernel bug reports coming from Borland
>while developing Kylix. You'll see what kind of support they got from the
>Linux community. Mostly "we don't care about a commercial vendor" kind of
>reponses.

Sound _very_ silly. _Any_ bug report is valuable, as it might improve the
software !
It has nothing to do with commercial or not, we have the same problem with
FPC when trying to interface to glibc. (or MySQL which is also a binary
compability hell).
Luckily with FPC we have several advantages:
- We are not as far as Borland yet, so less problems.
- An upgrade doesn't cost money, so we don't have as many versions to
support, people can locally adapt/apply fixes for {*word*118} distro's.
However simply put the problem is that Linux (and the other free *nixes too,
though to a lesser extent) are typically distributed as one consistent big
set of C (or more precise GCC and derivates) code. Backwards compability
only exists as C code, via a gcc compiler and headers.
(excluding a few simple pic and/or fully statically linked binaries)
The only typical Linux binary apps are acroread and (in former times)
Netscape, and they sport huge compat libraries and a fairly high amount
of static linking. Moreover they are not development packages, and thus
don't generate binaries themselves.
If you have to connect to that on any level that is not 100% the same
as the current gcc and header set, you are in for a lot of pain.
The problem is (in both cases) who should shift the work to reach that
binary compability? Borland or Linux?
Of course, most windows users will automatically say Linux, since Windows
binary compability is quite normal, even expected, but try to put yourself
into the shoes a Linux distro maintainer. Or a volunteer programmer that
must do this kind of boring work, while Borland reaps commercial profits
from it.
I think the main problem is that Borland is not big enough, so that it can
actively engage in the commercial side of Linux distro's (e.g. buy a distro
or choosse a few distro's) and appoint some fulltime maintainers to keep the
distro Borland friendly (read: binary compat).
Nearly all other large OS related firms bought or worked themselves into
some major part of the community. The idea that that everything is still
volunteer work is a myth:
IBM: Linux Kernel and enterprise kernel stuff
SUN: Java, Gnome, internationalisation, StarOffice/OpenOffice
Apple: BSD and binutils/gcc.
Novell: Suse, Mono, (and Samba IIRC), KDE via Suse
RH: binutils/gcc, kernel, own distro (sponsors Fedora maintenance)
Borland: ?
Quote
Now with Novell/Suse supporting the Mono project I suppose the relationship
between the programming tools partition of the Linux community and
commercial vendors seems to be improving.
Mono is unrelated to this problem set, since Mono is GCC based, will be
distributed with the distro.
So IMHO it is not a question of political will in the OSS community to
cooperate, but who is going to pick up the tab for the boring work that
Borland needs to be done on a reasonable timescale (even though I
immediately agree that it wouldn't hurt if Linux was a bit more backwards
compat. Actually one of the reasons that I use FreeBSD)
 

{smallsort}

Re:RE: Kylix now officialy dead?

Quote

Me? Or do you mean Simon, who wants to write a remote de{*word*81}.

I read Simon's post only after I wrote mine <g>.
-Michael
 

Re:RE: Kylix now officialy dead?

Quote
Great !

I really wonder how that can be done ! Please keep us posted.

Will it be usable from the Delphi IDE like when remotely debugging on a
Windows target, or will it use a different frontend ?
The plan is to get it natively integrated into the Delphi 2005 IDE, which
supports plugable de{*word*81}s, so it should be perfectly integrated there.
For Delphi 6/7, it probably will be integrated into the IDE, but not as
tight as for the Delphi 2005. Probably you will have to use separate
watches windows etc. Breakpoints etc should work using the standard
code editor.
I'm also investigating using the Remote Debugging protocoll from Borland,
but currently am unsure if that would work at all - I don't know which
parts of the work are done on the local, and which parts on the remote
side.
The main work to do actually won't be the de{*word*81}, but the evaluator.
Probably looking at some of the various pascal scripting engines might
help here. While my knowledge regarding RTTI ist quite ok, it's not
really extensive when it comes to the VMT etc.
Simon
 

Re:RE: Kylix now officialy dead?

Quote
I suppose Delphi 2005 has a remote de{*word*81}.
No, sadly it doesn't have one.
Quote
So the frontend is there and (provided we have the appropriate
compiler/linker/libraries) just the Linux debug runtime is needed to be
added to have a full "remote IDE".
No, it's not that easy. Remember the local part of the de{*word*81} is not able
to work with the Linux binary and its symbol tables etc.
There is a lot of stuff I still need to do research on.
Luckily Borland is supporting me.
Simon
 

Re:RE: Kylix now officialy dead?

Simon Kissel schrieb:
Quote
No, it's not that easy. Remember the local part of the de{*word*81} is not able
to work with the Linux binary and its symbol tables etc.

There is a lot of stuff I still need to do research on.
Did you ever think about using the gnu de{*word*81} for this project?
AFAIK it supports remote debugging via serial connection and cygwin
provides a windows port.. (Havn't tested it, though)
Quote
Luckily Borland is supporting me.
Good news.. Now any progress in the community thingy?
Willi
 

Re:RE: Kylix now officialy dead?

Quote
>No, it's not that easy. Remember the local part of the de{*word*81} is not able
>to work with the Linux binary and its symbol tables etc.
>
>There is a lot of stuff I still need to do research on.

Did you ever think about using the gnu de{*word*81} for this project?
AFAIK it supports remote debugging via serial connection and cygwin
provides a windows port.. (Havn't tested it, though)
I don't really like gdb, and I can't imagine using it for real work. Probably
I'm spoiled by the Delphi/Kylix de{*word*81} ;)
And I don't know if it would be possible to implement lots of Delphi
specifics in it. And there would be the license thing - I don't want
to infect the project with the GPL again.
I prefer doing this myself, as I have to learn the internals anyway
to get the Delphi-specifics to run...
Quote
>Luckily Borland is supporting me.

Good news.. Now any progress in the community thingy?
No, not much - the Borlanders seem still to be busy with Delphi 2005
and its first patch...
Simon
 

Re:RE: Kylix now officialy dead?

Simon Kissel schrieb:
Quote
I don't really like gdb, and I can't imagine using it for real work. Probably
I'm spoiled by the Delphi/Kylix de{*word*81} ;)
I know that feeling.. (Me too?)
I'm only touching gdb via ddd..
Quote
And I don't know if it would be possible to implement lots of Delphi
specifics in it. And there would be the license thing - I don't want
to infect the project with the GPL again.
Well, if you look at ddd, gdb is quite usable and if Kylix based
executables are compiled with stabS info, it should be possible to debug
them with gdb..
GPL is a problem if the de{*word*81} would have to be linked against the gdb
source. If gdb is just called in the background, it should be no issue.
But I've no experience in this field, so I'm just guessing..
Quote
I prefer doing this myself, as I have to learn the internals anyway
to get the Delphi-specifics to run...
I'm looking forward to see your Version 0.1 :-)
Quote
No, not much - the Borlanders seem still to be busy with Delphi 2005
and its first patch...
I sure hope this is no vicious circle..
Willi
 

Re:RE: Kylix now officialy dead?

On 2004-11-24, Willibald Krenn < XXXX@XXXXX.COM >wrote:
Quote
Simon Kissel schrieb:
>And I don't know if it would be possible to implement lots of Delphi
>specifics in it. And there would be the license thing - I don't want
>to infect the project with the GPL again.

Well, if you look at ddd, gdb is quite usable and if Kylix based
executables are compiled with stabS info, it should be possible to debug
them with gdb..
If you set the language to "pascal" your expressions can be more Pascal
like. (set language pascal in the console, this was added for FPC).
Ansistrings can be inspected with x/s
GDB is a bit harsh indeed, and I also prefer Delphi. However saying that
nothing can be accomplished with GDB is a bit simplistic.
 

Re:RE: Kylix now officialy dead?

Simon Kissel wrote:
Quote
If you are a Linux developer, who wants to develop only under Linux and
for Linux, forget about Kylix. Most "real Linux geeks" don't want RAD, but
{*word*155} low-level c. Near all of them don't want to work with
closed-source software, and even less want to pay money for software.
That's the reality Borland has faced when trying to sell Kylix.
Sorry for coming to the party late, but I'm one of those "wants to develop
under linux for linux" types, and I *really* do like "rad" environments. I
never had a problem "paying for a tool" [heck, I even bought
"objectwindows" if you remember that!]
Quote
I personally think that 99% of all people that paid for Kylix are Users
coming from the Delphi Windows world. From a business standpoint it would
totally make sense to keep these customers.
which, I guess, leaves me in the 1% minority that Borland can shit upon?
Quote
Personally I would not miss the Kylix IDE at all. I never used it. I
always created my Applications under Windows and then compiled them on
Linux. I have the feeling I'm not alone. The number of CrossKylix users
points into the same direction.
This is almost a straw-man argument: since it is rapidly becoming appearhent
that "the only way" to do things appears to be to use crosskylix, it
naturally follows that "the number of users points in the same
direction..."
Quote
My personal hope is that Borland will integrate everything needed to build
Linux applications into the Delphi IDE.
Elsewhere in this thread (in response to R.F. Pels, it appears), you also
mentioned:
Only if you are part of the "I want to develop FOR linux ON linux"-group.
Which seams to be rather small. Actually you are the first one that group
to speak up ;)
---------
That's only true if he said it prior to 2002 :) I've been saying this for
quite some time, but Borland's actions have spoken louder than my pleas, so
I've stopped talking long ago. In fact, around that time I was seriously
developing an arcade-style game [believe it or not] because I was otherwise
"unemployed", and figured I could make a few bucks by actually creating a
game "for linux users". Turns out the alpha test went rather well, but as
soon as I tried it on a different desktop [same vendor and release of
linux], all hell broke loose and the graphics simply "didn't line up"
anymore. After a short search in the newsgroups here, I found it was a
"known problem", and that borland had zero intention of producing a fix.
 

Re:RE: Kylix now officialy dead?

Tom,
Quote
>I personally think that 99% of all people that paid for Kylix are Users
>coming from the Delphi Windows world. From a business standpoint it would
>totally make sense to keep these customers.

which, I guess, leaves me in the 1% minority that Borland can shit upon?
No, of course not. But hey, you have to face the reality: There currently isn't
much activity regarding Kylix from Borland. The Kylix IDE is a dead-end, now
that Delphi 2005 uses the new Galileo IDE. Also the Kylix IDE Winelib bindings
have proven to be a major PITA. All in all this means: A Kylix 4 IDE that's on
par with the Windows offering probably would mean a rewrite and thus a pretty
big investment. I would really love to see this, but currently chances don't look
too good.
Compared to this, the "Develop on Windows and deploy to Linux"-route would
be far less of an investment, and probably will give a better revenue. And that's
what it's all about for Borland, I guess.
IOW: I'd also prefer to get a new Kylix IDE version. But if I have to chose between
"no Linux offering at all", and a Windows IDE that is able to deploy to Linux, the
latter one sure would get my vote ;)
Quote
This is almost a straw-man argument: since it is rapidly becoming appearhent
that "the only way" to do things appears to be to use crosskylix, it
naturally follows that "the number of users points in the same
direction..."
No, I don't think that's the case. All Kylix third-party stuff I've seen until today
seems to be coming from Windows guys. I think it's pretty obvious that Borlands
Delphi customer base is coming from DOS/Windows, and not from UNIX. It's no surprise
that most of those still develop under Windows.
Quote
That's only true if he said it prior to 2002 :) I've been saying this for
quite some time, but Borland's actions have spoken louder than my pleas, so
I've stopped talking long ago. In fact, around that time I was seriously
developing an arcade-style game [believe it or not] because I was otherwise
"unemployed", and figured I could make a few bucks by actually creating a
game "for linux users". Turns out the alpha test went rather well, but as
soon as I tried it on a different desktop [same vendor and release of
linux], all hell broke loose and the graphics simply "didn't line up"
anymore. After a short search in the newsgroups here, I found it was a
"known problem", and that borland had zero intention of producing a fix.
This sounds rather vague. Care to post the QC reports involved?
Simon
 

Re:RE: Kylix now officialy dead?

Quote
OK it's a sacrilege to use Windows to
develop for Linux, but that is the Borland way....
Oh, I'm sure there are some Linux zealots who think it's "sacrilege."
OTOH, I've seen a lot of interest in developing on Windows for deployment
in Linux, including outside the Delphi community. It was a major topic
of discussion at the last .NET User Group meeting I attended (where
Delphi users were definitely in the minority).
Rick Carter
XXXX@XXXXX.COM
Chair, Delphi/Paradox SIG, Cincinnati PC Users Group
--- posted by geoForum on delphi.newswhat.com
 

Re:RE: Kylix now officialy dead?

Simon Kissel wrote:
[previously, I wrote]
Quote
>[...] Turns out the alpha test went rather well, but
>as soon as I tried it on a different desktop [same vendor and release of
>linux], all hell broke loose and the graphics simply "didn't line up"
>anymore. After a short search in the newsgroups here, I found it was a
>"known problem", and that borland had zero intention of producing a fix.

This sounds rather vague. Care to post the QC reports involved?
Yes, I know it is vague -- in a nutshell, I had done the development work on
my laptop with a screen resolution of 1400x1050, and everything "fit
nicely". When I tried to install it on my desktop with a screen resolution
of 1280x1024, it didn't "fit" the screen anymore, and some things that
should have been displayed simply weren't displayed. When I asked about it
in these newsgroups, I was given an answer to the effect of "yeah, it's
weird, but there isn't much you can do about it".
Shortly after that, I became officially "employed", so diddling around with
the game wasn't a priority anymore, and I pretty much dropped it. I
suppose I could dig up the sources if anyone is truly interested (though
since I never did officially release it, it doesn't have all the proper GPL
"verbiage" attached to it, but come to think of it, I wasn't entirely sure
I was going to release it as GPL at the time...)
 

Re:RE: Kylix now officialy dead?

Arthur Hoornweg < XXXX@XXXXX.COM >writes:
Quote
A sad day. Delphi 2005 will have no more CLX support or
so I've heard.

So any future Kylix version will no more have the "platform
independence" feature because you can't recompile a Kylix
source under Delphi or vice versa.

So guys, it looks like Kylix is dead...
Join the club.
BCB is dead too.
Perhaps now the Kylix hangers-on will acknowledge reality and move on.