| Re: IDL 6.0 on Fedora Core 1 Linux [message #37038] |
Thu, 20 November 2003 11:10  |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Thu, 20 Nov 2003 11:16:16 -0700, Paul Hein wrote:
> Checking the rpm's, I find
> RH 7.3 glibc-2.2.5-43.i386.rpm
> RH 8 glibc-2.3.2-4.80.6.i386.rpm
> RH 9 glibc-2.3.2-27.9.i386.rpm
> FC 1 glibc-2.3.2-101.i386.rpm
>
> So that looks good, glibc-2.3.2 is used for RH 8, 9 and Fedora Core 1.
> The installation booklet says, "IDL 6.0 was built with Linux 2.4 kernel
> with glibc 2.2 using Red Hat Linux." Glibc 2.3.2 works and hopefully
> 2.3.2-101 will work too.
The reality is IDL has little exposure to Linux distribution internals
-- it's fairly self-contained. It does link to glibc, which is
certainly a point of exposure to the distribution, and the
loader/linker libs, which are very stable (by definition). It uses
gcc in a minor way with MAKE_DLM, but otherwise has little reliance on
the actual flavor of Linux it is being run on. I'd recommend that RSI
just defines the basic standards, e.g. glibc>2.3.2, gcc> 2.9.x,
ELF-binary format support, etc., and compiles a list of distributions
which meet those standards. It should be a fairly large list. The
support issues might be more problematic, but I think this is the
correct way forward. The Linux Standard Base project might also be
interesting: http://www.linuxbase.org/.
JD
|
|
|
|
|
|
|
|
| Re: IDL 6.0 on Fedora Core 1 Linux [message #37045 is a reply to message #37043] |
Thu, 20 November 2003 10:04  |
Paul Hein
Messages: 6
Registered: November 2003
|
Junior Member |
|
|
An "Informed source at RSI" emailed me and this is what he said:
It's a good question. We're trying to figure that out ourselves...
I can tell some things right now:
1) We may test with RedHat Enterprise, but we don't want to
make it our standard reference platform. As you say, it's
expensive. To my mind, you might as well run Solaris
for that kind of $$$.
2) Fedora is a possibility, but it remains to be seen if it
will be as solid as Red Hat was, and we're waiting along
with everyone else to see.
3) For a couple of days, I thought that Suse might be the
answer, but then they got bought, and there are even more
questions about their future than there are for RedHat/Fedora...
4) Debian is a possibilty, but they are very slow to issue
updates, and their installation process is painful. I know
both of these are changing, but change takes time, and who
knows....
RedHat served an important purpose in the Linux community, even for
those who run other distributions. By establishing a solid least common
denominator, they brought order to that world. Pity they couldn't keep
doing it...
So, it's up in the air right now. Fortunately, there's no huge rush
to decide, and we can wait a bit to see how things shake out.
Our next release will probably be based on one of the already existing
RedHats, maybe even staying at RedHat 7.1 (it seems solid, and we've had
little or no compatability issues with other distros). Beyond that, we'll
watch events unfold, and try to pick the best one...
Please feel free to share this information with the newsgroup. The
opinions of the IDL Linux community are very important in figuring
out what we should do...
|
|
|
|
| Re: IDL 6.0 on Fedora Core 1 Linux [message #37069 is a reply to message #37045] |
Tue, 25 November 2003 07:45 |
Karl Schultz
Messages: 341
Registered: October 1999
|
Senior Member |
|
|
"Paul Hein" <hein@olympic.atmos.colostate.edu> wrote in message
news:3fc28eef@news.ColoState.EDU...
> David Fanning wrote:
>
>> Paul Hein writes:
>>
>>
>>
>>> Unfortunately I got some floating point error messages.
>>>
>>>
>>
>> Alas, this appears to be standard operating procedure,
>> as least with respect to translating Matlab code.
>> Floating underflow and overflow messages are driving
>> me batty. :-)
>>
>>
> Sorry! I was not clear at all. With Fedora, the demo program crashed
> with a floating point error that the terminal window recorded. The 2
> programs, Object World and 3D
> Geometry Demo, (I think those were the ones) crashed on start up leaving a
> floating point error message. They did not crash the idldemo program.
Perhaps it would help if you could post the actual text of the error
messages. That might make it possible to figure out if IDL or some other
component in the system is issuing the message.
As already mentioned in this thread, another potential source of instability
is the X11 server. I've got no idea at the moment how good the X11 code is
in Fedora. I know that the Redhat folks did put a lot of effort into X11,
including the packaging and configuration to support hardware-accelerated
OpenGL for a pretty good range of hardware. The hardware-accelerated OpenGL
stuff can be pretty touchy at times, so it wouldn't be too unlikely for it
to be part of the problem.
Try running some other OpenGL client programs. The most likely one you'll
find in your distro is glxgears. Also look for a game like tuxracer. If
glxgears throws errors or crashes, then the problem lies in your OpenGL
support. You might consider changing your XF86config file to avoid using
the hardware driver, or get an updated hardware driver from your video card
vendor. It is not always trivial to do this update because it sometimes
involves building a kernel driver. But the bigger card vendors like nVidia
and ATI provide pretty good support in this area.
If you don't want to mess with your X server configuration, you can try
disabling hardware-accelerated OpenGL graphics in IDL. One way to do this
is to put "idl.renderer: 1" in your X resources.
There is a lot more information in the RSI Tech Tip at:
http://www.rsinc.com/services/techtip.asp?ttid=3524
Hope this helps,
Karl
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Re: IDL 6.0 on Fedora Core 1 Linux [message #37111 is a reply to message #37038] |
Fri, 21 November 2003 09:50 |
George N. White III
Messages: 56
Registered: September 2000
|
Member |
|
|
On Thu, 20 Nov 2003, JD Smith wrote:
> The reality is IDL has little exposure to Linux distribution internals
> -- it's fairly self-contained.
Some IDL apps use the X-server, which in my experience has been the
biggest source of instability, mainly because new systems tend to come
with hardware that is too new for the software to be fully debugged, but
I gather that some cheap 2rd party boards using name-brand chips cheat
on certain specs to cut costs and are simply not up to demanding
applications.
> It does link to glibc, which is
> certainly a point of exposure to the distribution, and the
> loader/linker libs, which are very stable (by definition). It uses
> gcc in a minor way with MAKE_DLM, but otherwise has little reliance on
> the actual flavor of Linux it is being run on. I'd recommend that RSI
> just defines the basic standards, e.g. glibc>2.3.2, gcc> 2.9.x,
> ELF-binary format support, etc., and compiles a list of distributions
> which meet those standards. It should be a fairly large list. The
> support issues might be more problematic, but I think this is the
> correct way forward. The Linux Standard Base project might also be
> interesting: http://www.linuxbase.org/.
Within a given distro, not all video hardware has robust X servers.
Would the support issues be reduced if, in addition to the above
basic standards, RSI compiled a list of support graphics hardware
and X servers?
If companies like RSI find it too hard to support all linux versions
that satisify certain basic standards, then it is time for the academic
community to organize a distribution targetted at the needs of
science and specifically tuned to serve as a platform for IDL, matlab,
commercial compilers, data aquisition hardware, etc. This could be
handled by contracting with a commercial vendor under the guidance of
a working group with reps from the above vendors and the user community.
I can envison the day when such a linux distro is part of your society
membership (and included on the DVD that contains the monthly newsletter).
It would be much simpler if we all went out and purchased Apple machines.
--
George N. White III <aa056@chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia, Canada
|
|
|
|
|
|
|
|
|
|
| Re: IDL 6.0 on Fedora Core 1 Linux [message #37217 is a reply to message #37105] |
Tue, 25 November 2003 13:51 |
R.Bauer
Messages: 1424
Registered: November 1998
|
Senior Member |
|
|
Mirko Vukovic wrote:
> Reimar Bauer wrote:
>
>> Mirko Vukovic wrote:
>>
>>> Paul Hein wrote:
>>>
>>>> Has any one yet tried out Fedora Core 1 (RedHat 10's new name) Linux
>>>> with IDL? I was wondering if it worked. What is RSI going to do
>>>> with their Linux versions? Are they going to switch to testing with
>>>> another linux distribution, or test with the expensive RedHat
>>>> Enterprise versions, or with the new Fedora versions? I am thinking
>>>> of upgrading from RedHat 7.3 and RedHat somewhat dropping support
>>>> for the free versions have me wondering what I should do?
>>>> Paul
>>>>
>>>
>>> It works for me on Debian Woody distribution on my 180MHz box. It
>>> installed without a hitch.
>>>
>>> I am trying to see what is the minimal hardware config for IDL these
>>> days.
>>>
>>> Mirko
>>>
>>
>> I am interested in this too please do a posting afterwards.
>>
>> Reimar
>>
>>
>
> Reimar,
>
> what exactly did you have in mind? My message was intented as a bit of
> a joke, as my installation is probably one of the least powerfull on the
> planet.
>
> But do you need some performance specs for it or something else (like
> overall ease of use and IDL speed and responsiveness)?
>
> Mirko
>
Dear Mirko,
I am woking at the moment on idl in a UML. The UML could be used with
different setups. One is to use it without X-SERVER and doing all in the
Z-Buffer or using redirection of the display to a different machine.
The reason for this is. I like to call idl by a wiki prozessor and this
protects the apache server to be killed. Otherwise a mistake like
file_delete,file_search() will clean the whole box.
If the UMl in this way is killed it is easy to restore because its only
one file to restore.
By always much memory I have not thought about the real necessary memory
which idl needs to run.
Reimar
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
