Board index » kylix » Re: What to do with Kylix

2004-02-14 03:08:56 AM
kylix1
Hrvoje Brozovic wrote:
problem you may face is in the somewhat cumbersome debugging of the
CGI. So before you even start any CGI project that needs to be
deployed into remote hosted website, you probably will do yourself a
favor and make for yourself first a dedicated ftp-client which could
upload automatically your CGI from your local machine in to the remote
site, and make it executable and set or disable other permissions.
But probably the best solution is to setup a separate Linux box and
install on it Apache recreate all the directory structure you have in
the remote site and use this box for testing.
juliusz
--
InstallMade - Kylix-specific installer
www.superobject.com/installmade/
www.superobject.com/imoe/download.html
 

Bob Swart schrieb:
 

"Iman L Crawford" <ilcrawford.at.hotmail.dot.com>wrote in message
"But the open source revolution has yet to catch on at the corporate
desktop. Linux represents only 2.8 percent of the client operating system
market, according to IDC analyst Dan Kusnetzky. IDC doesn't expect that
figure to change significantly over the next three years. "
 

{smallsort}

Bob Swart wrote:
about Borland NOT releasing any source code for their Xerces/Xalan
DOM-wrappers?.
see:
<groups.google.com/groups&lr=&ie=UTF-8&selm=3d8a65b0%241%40newsgroups.borland.com>
I am talking about being stuck to the old version 1.6.something.
I already had to do a dirty hack[1] to get it work under SuSE 8.2, and by
trying SuSE 9.0 - eval, I expect more trouble, since I saw more errors.
[1] Editing the binaries with HEX - editor.
--
Best regards
Stig Johansen
 

Hi Stig,
6.x/7, and something like SuSE 9 for deployment).
Bob Swart (aka Dr.Bob - www.DrBob42.com)
--
Bob Swart Training & Consultancy (eBob42) Borland Technology Partner
webmaster UK-BUG / DDG Developers Group - IntraWeb Authorized Trainer
 

JQP wrote:
COUNT the number of machines sold with Linux and the number of Linux copies
purchased. Anyone who knows ANYTHING AT ALL, about the way Linux is used
and installed, knows that using these types of figures as a way to measure
use, IS JUST PLAIN DUMB.. I have given you previously, one small instance,
our own Consulting company. We buy one Linux copy, and run it on ten
machines. MS came on three of those machines, and it runs on ZERO of those
machines. So according to the survey our company consist of the following:
3 Windows Desktops
1 Linux Desktop.
1 Windows Server
0 Linux Servers
The true facts are:
1 Windows Server
1 Windows Desktop
7 Suse Linux Desktops
3 Suse Linux servers
Doesn't jive with the sales figures, now does it?
Again the very FALSE impression is that all those running Linux buy a new
copy for each machine they own and/or every machine they buy comes
preloaded with Linux. The truth of the matter is, I do not know of a single
solitary company, nor individual, where this statement would be true, so
the statistics are DEAD WRONG.
Taking then into account, our one small company, that means the the IDC
figures are actually 10 times in error in our company alone. Now suppose
that we even half that figure to say 5 times wrong. That would means that
Linux desktops could be at 50% error (100% in our case) 2.8X5=14% of
desktops running Linux. We have not even taken into account free
downloadable linux versions like Mandrake ISOs, Debian ISOs, Gentoo,
Slackware, etc.
There is, in actuality, no way to measure the Linux Market penetration at
all. The Netcraft survey for servers, would be closer, to be sure, but even
then, you would have to assume that all Linux servers were connected
directly to the internet. In our case, that would not be true, and I doubt
that ALL internal servers used in households, or corporations would all be
tied to the internet either. It would not be true of MS servers either. MS
servers are easier to track because if you do not buy a license, MS can
legally come after you.
This survey must also assume, that everyone who buys a machine with Windows
preinstalled, still runs Windows. I just provided you with an example of
three machines in our company alone, where this is not true. Why not ask
the users here who bought a machine with Windows on it, If they still run
Windows on that machine? For each one who states, "NO", then that means
MS=-1 and whatever (BSD, Linux, Solaris, whatever)=+1
I can also give you one of our smaller customers as an example. They always
buy Dell machines, so they all come pre-installed with Windows. Their last
purchase was 25 machines. 15 of these run Windows, the rest run Linux. They
did not buy (surprise, surprise) 10 Linux license. In fact, they are
runing Gentoo and Fedora, so they have not purchased a single Linux
license. Therefore, according to IDC, they are running:
25 machines on Windows
0 machines on Linux.
The truth is:
15 machines on Windows
10 Machines on Linux
Meaning that the IDC figures are now off by 40% for Windows, and (since they
are showing zero running on Linux) 10 times 1, wrong about Linux.
Again, this is an example of only two companies. Already you have a deficit
of 9+10 machines WRONGLY counted in this survey or 19 out of 37 machines
calculated incorrectly. Those are pretty large numbers of a percentage
wrongly counted to come up with their figures. And our two companies are
certainly not alone.
In fact three of our main customer's customers replaced their Windows Server
installations with Linux last week. They would still be considered in the
Windows camp, albeit they are running Linux. No, this is not a desktop
case, but the same IDC figures take into account servers as well. A trip to
Netcraft alone, would reveal that their Linux Server figures cannot be
anywhere near correct either. I would doubt either Oralce, IBM, or Sun, for
instance, purchase a brand new copy of Linux for each internal machine they
use at either the Desktop or Server level. In fact, IDC probably still
shows the majority of Oracle Corps. Desktops running Windows, which is DEAD
WRONG.
 

"pnichols" < XXXX@XXXXX.COM >wrote in message
industry; including Linux distro vendors; accept the IDC survey as the best
possible numbers. In particular, I recall reading an interview with a Suse
representative where he said as much.
In any case, do you have better numbers or do you operate solely on
rhetoric?
 

Bob Swart wrote:
Yes, presumably because you don't use xerces/xalan to do XML/XSLT
processing.
(Notice, this is *also* a deployment problem)
The problem (unresolved externals) will occur with newer versions of GLIBC.
The first problem you will encounter is with GLIBC 2.3.2 introduced with
SuSE 8.2 and AFAIK RH9 is the __pthread_atfork problem.
You can verify this by issuing:
ldd-r libxercesxsldom.so.1
I've posted a workaround to QC some time ago.
As I wrote, I tried a SuSE 9.0 eval to see if the problem still exists, and
it did, and one more.
If anyone have a SuSE 9.0, I would be gratefull, if you could post the
result of the above command.
(I've forgotten the output)
--
Best regards
Stig Johansen
 

pnichols wrote:
network.
It seems to be pretty popular here in dk for people with broadband/ADSL.
--
Best regards
Stig Johansen
 

Bob Swart wrote:
Did you try to deploy apache dso for Redhat 9? Everything is fine for
Redhat 7, 8. However, it always caused segment fault when apache
loading the dso.
Alice

Hi Alice,
(kernel 2.4.23) on the deployment server. Haven't heard of any problems,
yet ;-)
Bob Swart (aka Dr.Bob - www.DrBob42.com)
--
Bob Swart Training & Consultancy (eBob42) Borland Technology Partner
webmaster UK-BUG / DDG Developers Group - IntraWeb Authorized Trainer
 

JQP wrote:
us, they have not contacted any of our customers who have or who are
switching a large segment of their Windows servers and desktops to Linux.
So, if we are all being ignored, then the accuracy of these estimates,
cannot be accurate.
Would be intersting to here from the site alone, how many are running Linux
and how many have been contacted by IDC. Bet it will be zero to maybe two
that have been contacted by IDC. So that in and of itself, means that an
accurate reflection is impossible.
free operating system usage?
I gave you an two different companies, neither of which has been contacted
by IDC, where the estimated processes used for calcualtion pruposes are
extremely inaccurate, to say the least. And would be the only two in the
entire world? I think not.
It is much easier to estimate commercial proprietary products like Windows,
Solairs, AIX, etc. than it is to measure Open Source like Linux, Free BSD,
ect. How would you track these? You could use downloads from all sites who
host these distros, and that would be more accurate, but even then, you
would have to preclude the following facts:
(1) That each user that did download these distros and installs, also
continued to use them excusively.
(2) That each downloaded copy was downloaded separately for each
installation (we know that this would be totally wrong).
So any effort to esitmate usage is at best an educated guess, and at worst,
totally inaccurate. I do not know of any methodology that would accurately
track Linux usage overall. I do understand the intent, but I do not know
how you can give an accurate protrayal.
 

pnichols wrote:
news web-site. They tried to identify OS from browser identification
string. The numbers was something like
1. different windowses share 90% of all hits
2. linux about 5%
3. other os-es share another 5%
Ok you may say that many Linux browsers try to pretend to be MSIE. However,
Konqueror by default is not pretending, Mozilla also, Opera pretending
being used under Windows but not under Linux.
It is possibly that IDC gather statistic in similar way.
 

Ender wrote:
statistic, the IDC report would be off by 2.3%.
This would, of course, assume (even if you measured by this) that all Linux
desktops are connected to the internet. Agreed, that this would be closer
to the truth than measuring by sales of computers or the OS, however. This,
in fact, would be a much closer representation, by capturing the header
files as per OS. In Windows, it would work (not sure about MACs, since like
in Konqueror, you can suppress sending this information, as per
Linux/Unix). But it would be a much better representation of actual use.
It is also correct that you can set up the browser to report as IE. Why
would some wish to do this? Perhaps because some sites do not support other
browsers. I have got this option to work sometimes in Konqueror, and also
in Opera. IE mimicking seems to work better in Opera than anything else I
have tried, but I mainly use Mozilla. So do a lot of Windows users, but it
would still report the OS, unless they turned this option off (can you do
this in Mozilla Windows, I haven't had a Windows machine near the internet
in over a year :)).
 

something like this: Win: 95%, Linux: 4%, Others: %1 and so on... or Win:
93, Linux: 3, Others: 4... so initial numbers was not precise. I remember
than Windowses claim more than 90%, and other OS-es was roughly equal with
Linux.
under Linux and even can't find that setting. On the lunch i'm asked the
colleagues who changed identification strings of browser (they all Linux
programmers)? It was 1 from 13.
Additionally about servers.... measuring server count also cannot be
accurate. For example in my company there is plethora of Linux servers, i
can't count them. However each of the servers really doing one kind of
work. We getting some rusty box installing linux on them and forgetting
somewhere on the shelf, where it running single app usually with 1% of
load. At the same time there are few heavyloaded windows machines each
running several apps. Of course we have heavyloaded Linux machines also.