Having used 6.1 for over three weeks now, I can confidently say it is a
big improvement in terms of stability over 6.0PB. Upgrading is still a
nightmare however (correct permissions would do a lot to alleviate
this). Also, I still have problems with dlopen (which used to work in
6.0 and 6.0PA but hasn’t worked since 6.0PB), and I would love to have
buildable sources for libc for 6.1 from cvs.qnx.com so that I could
investigate.
Anyway; overall a good job.
Rennie
Hmm. We had dlopen working in a small test program under 6.1
Rennie Allen wrote:
Having used 6.1 for over three weeks now, I can confidently say it is a
big improvement in terms of stability over 6.0PB. Upgrading is still a
nightmare however (correct permissions would do a lot to alleviate
this). Also, I still have problems with dlopen (which used to work in
6.0 and 6.0PA but hasn’t worked since 6.0PB), and I would love to have
buildable sources for libc for 6.1 from cvs.qnx.com so that I could
investigate.Anyway; overall a good job.
Rennie
Dean Douthat <ddouthat@faac.com> writes:
| Hmm. We had dlopen working in a small test program under 6.1
|
| Rennie Allen wrote:
|
| > Having used 6.1 for over three weeks now, I can confidently say it is a
| > big improvement in terms of stability over 6.0PB. Upgrading is still a
| > nightmare however (correct permissions would do a lot to alleviate
| > this). Also, I still have problems with dlopen (which used to work in
| > 6.0 and 6.0PA but hasn’t worked since 6.0PB), and I would love to have
| > buildable sources for libc for 6.1 from cvs.qnx.com so that I could
| > investigate.
I found that dso’s I built under 6.0 that worked fine with dlopen
had to be rebuilt under 6.1 before I could get dlopen to work (otherwise
it died an ugly death in the ldqnx code in libc).
Similarly, I could build with either qcc or gcc in 6.0, but in 6.1 I have
to build with qcc to get it to work, at least some of the time.
I tried to narrow down a test case, and just how qcc was doing things
differently, when all of a sudden (no reboots, etc.), the gcc built objects
also worked under 6.1.
Anyway, I’d definitely try completely rebuilding the dynamic objects from
scratch with qcc, and see if that helps…
I can get dlopen to work fine when calling it from C source, my problem
is with doing dlopen on a shared object created from C source, from a
C++ program. This used to work fine with 6.0a. I have rebuilt
everything many times. The C++ program calling dlopen ends up running
ready inside dlopen. If I could build the source for 6.1 I could find
out exactly what is going on.
-----Original Message-----
From: Dave Olson [mailto:olson@free.tahoenetworks.com]
Posted At: Thursday, August 02, 2001 6:44 PM
Posted To: os
Conversation: Kudos on 6.1
Subject: Re: Kudos on 6.1
Dean Douthat <ddouthat@faac.com> writes:
| Hmm. We had dlopen working in a small test program under 6.1
|
| Rennie Allen wrote:
|
| > Having used 6.1 for over three weeks now, I can confidently say it
is a
| > big improvement in terms of stability over 6.0PB. Upgrading is
still a
| > nightmare however (correct permissions would do a lot to alleviate
| > this). Also, I still have problems with dlopen (which used to work
in
| > 6.0 and 6.0PA but hasn’t worked since 6.0PB), and I would love to
have
| > buildable sources for libc for 6.1 from cvs.qnx.com so that I could
| > investigate.
I found that dso’s I built under 6.0 that worked fine with dlopen
had to be rebuilt under 6.1 before I could get dlopen to work (otherwise
it died an ugly death in the ldqnx code in libc).
Similarly, I could build with either qcc or gcc in 6.0, but in 6.1 I
have
to build with qcc to get it to work, at least some of the time.
I tried to narrow down a test case, and just how qcc was doing things
differently, when all of a sudden (no reboots, etc.), the gcc built
objects
also worked under 6.1.
Anyway, I’d definitely try completely rebuilding the dynamic objects
from
scratch with qcc, and see if that helps…
Hi Rennie,
Could you send me a reproducable test case where you see this happening
so I can test it on my 6.1 machine. Any additional info that you think
would be relevant would also be appreciated.
Thanks.
Barry Faubert
bfaubert@qnx.com
QNX Software Systems Ltd.
Rennie Allen <RAllen@csical.com> wrote:
I can get dlopen to work fine when calling it from C source, my problem
is with doing dlopen on a shared object created from C source, from a
C++ program. This used to work fine with 6.0a. I have rebuilt
everything many times. The C++ program calling dlopen ends up running
ready inside dlopen. If I could build the source for 6.1 I could find
out exactly what is going on.-----Original Message-----
From: Dave Olson [mailto:> olson@free.tahoenetworks.com> ]
Posted At: Thursday, August 02, 2001 6:44 PM
Posted To: os
Conversation: Kudos on 6.1
Subject: Re: Kudos on 6.1
Dean Douthat <> ddouthat@faac.com> > writes:| Hmm. We had dlopen working in a small test program under 6.1
|
| Rennie Allen wrote:
|
| > Having used 6.1 for over three weeks now, I can confidently say it
is a
| > big improvement in terms of stability over 6.0PB. Upgrading is
still a
| > nightmare however (correct permissions would do a lot to alleviate
| > this). Also, I still have problems with dlopen (which used to work
in
| > 6.0 and 6.0PA but hasn’t worked since 6.0PB), and I would love to
have
| > buildable sources for libc for 6.1 from cvs.qnx.com so that I could
| > investigate.I found that dso’s I built under 6.0 that worked fine with dlopen
had to be rebuilt under 6.1 before I could get dlopen to work (otherwise
it died an ugly death in the ldqnx code in libc).Similarly, I could build with either qcc or gcc in 6.0, but in 6.1 I
have
to build with qcc to get it to work, at least some of the time.I tried to narrow down a test case, and just how qcc was doing things
differently, when all of a sudden (no reboots, etc.), the gcc built
objects
also worked under 6.1.Anyway, I’d definitely try completely rebuilding the dynamic objects
from
scratch with qcc, and see if that helps…