Board index » kylix » Re: So, what to use if Kylix is dead?

2003-11-05 12:53:07 AM
kylix2
On 03-Nov-03, JQP said:
--
Bill
--------
"We may become the first society destroyed by its own experts --
especially experts in fields where there is no expertise that can be
verified by facts." -- Thomas Sowell
 

On 03-Nov-03, Trane Francks said:
between Linux and commercial software. When Linux becomes viable as a
commercial platform, I'll be happy to use it. My only interest in an OS
is as a tool, a foundation on which to build. To use Linux today, I'd
have to either constrain my products to a narrow range of distros, or
resign myself to spending most of my time exploring the idiosyncracies
of the systems on which customers may choose to install. And so far,
the Linux docs still make Windows docs look pretty good.
--
Bill
--------
"We may become the first society destroyed by its own experts --
especially experts in fields where there is no expertise that can be
verified by facts." -- Thomas Sowell
 

platform. Yes he may lose his support, but often problem does not related
directrly to distro. In fact support engineers often not ask about what
distro is used.
 

{smallsort}

"William Meyer" < XXXX@XXXXX.COM >wrote in
ever done. I've never worked in the software industry though.
--
Iman
 

WM>have the luxury of living with our parents, in a rent-free space.
WM>The ideology of open source is fraught with peril for software
WM>professionals.
Ideology of open source is not better than other technologies in common but
it perfect from technical POV. Once you going to gather pearls on open
sourced shelf, you should be ready at least support your product as if it
was released as opensourced. There are companies that operate successfully
in Linux world. Borland moved to Linux with Kylix using purely Windows-like
support & deployment strategy. In Windows world that strategy would work
fine, but in Linux world it failed.
Many people point on that fact from the early times, even before Kylix
release. What they can do if Borland remains completely deaf to the voice of
reason? It seems that their ears opened only to the sounds of gold in the
pockets.
 

You cannot fight a crowd. Become part of it or leader of it or fall dead.
 

On 11/05/03 01:53 +0900, William Meyer wrote:
software on RH, SuSE, NT, 2k, XP, Solaris, HP-UX and IRIX. I have
a grasp of the issues involved. A firm grasp.
already a viable commercial platform.
you're not a Kylix programmer and possibly not even a Linux user?
on NT, 2k and XP and find differences in behaviour among all of
them. I've seen service packs break software, so our
installations are mated to a specific service pack revision.
As for releasing on a narrow range of distros, that's precisely
what we do for our Linux products. That's no different than a
Windows house releasing software for W2k and not supporting it if
the customer chooses to install it on XP.
discussion of programming with Kylix. The quality of a Kylix
app's docs have nothing whatsoever to do with Linux except in the
most incidental sense.
trane
--
//------------------------------------------------------------
// Trane Francks XXXX@XXXXX.COM Tokyo, Japan
// Practice random kindness and senseless acts of beauty.
// mp3.com/trane_francks/
 

On 04-Nov-03, Trane Francks said:
panacea, or at least, an OS suitable for all occasions, and that's just
not so.
hoping that someday, the situation will improve sufficiently that I can
make use of the environment.
expertise to survive Windows, and cannot (as yet) see sufficient ROI
potential to do the same for Linux.
minor detail. Of course, many (perhaps most) of those are firmly
entrenched in RH (now announced to be out of production!) In the
Windows environment, there are not so many idiosyncracies at the app
level, though there are certainly many mine fields for the device
driver folks.
that the docs are *not* a tragedy. For my part, I consider that the
Windows docs have always been a tragedy, and standards have simply
fallen to their level. But Linux docs have yet to attain such heights.
And too often, any rational discussion of Linux pro and con is
foreshortened by screams of "TROLL".
--
Bill
--------
"We may become the first society destroyed by its own experts --
especially experts in fields where there is no expertise that can be
verified by facts." -- Thomas Sowell
 

 

"William Meyer" < XXXX@XXXXX.COM >wrote in message
"I don't like what you're saying but I can't really refute it".
 

JQP wrote:
--
.. P.
 

JQP wrote:
(2) glib version only change about every major release cycle in the KDE and
Gnome dev cycles. No more the Windows dlls would change. But since Windows,
does not have any built in dev or app tools, I guess you can be safe paying
for them at infinitum, for simple upgrades. Fortunately, Linux elliminates
this need.
product line.
Apps written for one Linux platform will perform on all but the most
estranged Linux distro, provided the apps are statically linked. I
explained this to you once before JQP.
Think in terms of writing a Delphi app where you do not include your
packages in the compile. If you do not supply the proper Borland bpls with
your app, or do not insure that the proper bpls are installed on the
user's machine, you will have problems. It amounts to the very same thing.
Most Linux applications do not build statically linked rpms. Instead, rather
than deivering a 20 meg download, they deliver a 2-5 meg download, using
the library files that are, or should be, on each machine. Now if you want
to argue about consistency in making sure that each distributed program
offers a statically linked verses non statically linked app, we might find
agreement.
Apt-Get in the Debian distros, is a good example of how to distribute non
statically linked applications and at the same time, manage a way to find
all dependeny files that are not currenltly installed on the user's
workstation or server(s). Works well and more standards along this line are
in the work.
However in any situation, any entity that distributes any application,
should ensure that their applications are distrbuted with the necessary
support. If I ship a Delphi app on Windows, that I did not statically link
at compile time, then it is my responsibility that I give a path to the
user to make certain that al needed packages or dlls are also distributed,
with the correct version of those bpls or dlls. Again there is absolutely
no difference in this concept and the glib concept.
If borland wants to use the glibs on the Linux system, but does not bind
these libraries for the newer libs on newer systems, then that is not
Linux's fault, but rather Borlands.
Borland could remedy this problem by providing updates in terms of a make
file, or a self install file to check for the update to the newer glibs,
This would quickly remedy the QT libs Borland is using;-- it is the same
tactic QT itself uses. The only area where there may be problems, is in
using the proprietary Borland compiler and linkers. I do not know how these
are bound to the OS or to the Processor itself, as well as any HAL layers,
etc. This could be a problem for Borland in updating Kylix or it may not be
a problem. Since it is proprietary technology in Kylix, how could we know?
BTW, this is another reason I am thrilled that Borland decided to support
many compilers in Builder X. In so doing, we can use the gnu compilers and
link our own applications to various glib versions with our own
applications, without having to wait on an update by Borland.
 

On 11/05/03 11:46 +0900, William Meyer wrote:
weaknesses. I would say that "an OS suitable for all occasions"
simply does not exist. I haven't used an OS yet in which I didn't
run into some trouble.
for a specific distro with a disclaimer that it may work on other
distributions but is not supported. In our case, we're lucky
enough to have a niche product so we can dictate to the customer
what OS and version satisfies our system requirements. As a
concession, we often sell maintenance services along with the
product. As long as the customer maintains the system to meet our
requirements, we'll send an engineer to do periodic system
maintenance on the installed system. Of course, I realize that's
not a possibility for most software-only houses. We do both
hardware and software in a very small market, so we almost always
have support staff at the customer site anyway.
operating systems. Yes, I know it's not true by textbook
definition, but it can be as complicated. When you can have
different file system layouts, libs, supporting apps and even
kernel versions, it's the same as targeting HP-UX and Solaris.
man pages, HOW-TOs and newsgroups, real docs become more of a
"what's that?" question. I've been using Linux for about 8 years
now and I can honestly say that although I recognize the
weaknesses in the documentation versus other operating systems,
it doesn't actually bother me. I'm also in full agreement with
those who are quick to say that "if you don't like the docs, help
write better ones." It's a great way to give back to the community.
wind up speared by trolls in the end anyway. And there's LOTS of
trolling here, IMO. Pretty much any and every "Kylix is dead" or
"Borland killed Kylix" post could be considered a troll because
it is ultimately predictable and is designed to attract
predictable responses and flames. And when I see folks patently
stating that Linux is not a worthwhile platform to release
commercial software, I consider that to be a troll, too, simply
because companies have been targeting multiple UNIX flavours for
years without so much as a hiccough.
I can agree that Linux may not offer an acceptable ROI for a
particular product, but it just doesn't wash to categorically
discount Linux as viable for commercial releases.
trane
--
//------------------------------------------------------------
// Trane Francks XXXX@XXXXX.COM Tokyo, Japan
// Practice random kindness and senseless acts of beauty.
// mp3.com/trane_francks/
 

Oliver Feins wrote:
JSuite is pretty exhaustive, as the screen shots will indicate, but
expensive.
Many of the grids, etc. that require much custom programming in Delphi (due
to the compelxity of the WinAPIs), are much easier in Java, so you may not
have to buy third party controls. Using a TableModel, for instance, may
make it quite easy for you to roll your own custom Tables and Grids.
be found on Sourceforge (JFreeReport and DataVision). I generally create my
own reports using XML, so I have only rarely used these. JReport used to
ship with JB Enterprise, so I tried it out and it seemed to work quite
well. I will be evaluating JFreeReport because I have heard very good
things about it and it is FREE.
out there for Swing.
Since JB X will offer a free distrbutables version, that might fit your
bill. However Oracle JDeveloper is very good as well. The problem I have
had with JDeveloper is that it is closely bound to Oracle 9I, but if you do
not mind not having a plethora of Wizards at your disposal for generic type
programming, it works quite well. Oracle !0G however, looks extremely
promising.
I am not wild about Eclipse, but for SWT, it is the way to go. Net Beans is
good as well. For GUI development, Net Beans and JBuilder are close. For
any EJB or Enterprise development, I like JBuilder best for its more
independent support nature better, not to mention the toolsets.
The only downside to GUI development with NET Beans, is the STUPID design to
write certain code blcoks for you, that you cannot change in the IDE.
Sometime, I like to use my own defined functionality at certain init()
levels.
 

On 04-Nov-03, JQP said:
discussion.
--
Bill
--------
"We may become the first society destroyed by its own experts --
especially experts in fields where there is no expertise that can be
verified by facts." -- Thomas Sowell