David,
it's in the mail.
René.
On 22 sep. 2014, at 18:36, David Ashley wrote:
> Rene -
>
> Please send a copy of the Mac version of pthreads.h so I can resolve the
> problems.
>
> David Ashley
> w dot david dot ashley at gmail dot com
>
> On Mon, 2014-09-22 at 07:29 -0400, Rick McGuire wrot
Rene -
Please send a copy of the Mac version of pthreads.h so I can resolve the
problems.
David Ashley
w dot david dot ashley at gmail dot com
On Mon, 2014-09-22 at 07:29 -0400, Rick McGuire wrote:
> I changed the mechanism used to check for these symbols. The change
> worked on Linux, but I do
I changed the mechanism used to check for these symbols. The change worked
on Linux, but I don't know if this will fix your problem or not.
Rick
On Mon, Sep 22, 2014 at 6:27 AM, René Jansen wrote:
> done that. Yes. there is a difference.
>
> This is from the Cmake-generate new config.h :
>
> /
David will need to get involved with figuring out why the define is not
getting set correctly. That's his bit. Having all of those off is
definitely bad news, since it means the semaphores don't work (as you've
already discovered).
I just checked in some fixes for the CATD problems, but you shou
showing the differences:
420.txt
new.txt
OLD-> While the 4.2.0 version had:
NEW-> New
/* Define to 1 if you have pthread_mutexattr_settype function */
OLD-> #define HAVE_PTHREAD_MUTEXATTR_SETTYPE 1
NEW-> /* #undef HAVE_PTHREAD_MUTEXATTR_SETTYPE */
/* Define to 1 if PTHREAD_
done that. Yes. there is a difference.
This is from the Cmake-generate new config.h :
/* Define to 1 if you have pthread_mutexattr_settype function */
/* #undef HAVE_PTHREAD_MUTEXATTR_SETTYPE */
/* Define to 1 if PTHREAD_MUTEX_ERRORCHECK is a valid value */
/* #undef HAVE_PTHREAD_MUTEX_ERRORCHEC
One thing David used to figure out this problem on Linux was to build the
4.2.0 version (which uses the old autotools build) and compare the
generated config.h from that build to the one generated by cmake. The bit
of code that was messed up was the create method in SysSemaphore.cpp.,
specifically
yes, this is all pointing to problems with config.h,
Rick
On Sun, Sep 21, 2014 at 7:35 PM, René Jansen wrote:
> well, it did not solve the below problem, it did not overwrite my
> commenting out of the redefinition of semun.
>
> this is the stacktrace:
>
> Sampling process 1660 for 3 seconds wi
well, it did not solve the below problem, it did not overwrite my commenting
out of the redefinition of semun.
this is the stacktrace:
Sampling process 1660 for 3 seconds with 1 millisecond of run time between
samples
Sampling completed, processing symbols...
Analysis of sampling rexximage (pid
You'll probably need to debug and figure out where this is either hung or
looping and get a stack traceback if you can. rexximage hung on linux when
we were first converting to cmake and the problem ended up being issues
with the config.h file not setting the right defines. If it appears hung
in
This works. including fixing another problem
/Users/rvjansen/apps/oorexx/oorexx-code-0/extensions/rexxutil/platform/unix/rexxutil.cpp:247:7:
error:
redefinition of 'semun'
union semun
Which is also fixed with the latest level.
but: the build now seems to be stuck in the generation of re
I just checked in a tweak to TimeSupport to use the defines created in
config.h rather than the AIX define for deciding which header to include.
If this still fails, there might be a problem with how the config.h file is
getting built on the Mac. The constructed config.h file can be found in
the r
I think you're going to need to investigate this one from your end. I
suspect there are some setup problems in the config.h generated by the
CMake build. This code was not changed from the original, so it should
compile. The first thing you need to check is to see if maybe the AIX
#define is som
It did, but further down there is a new problem:
[ 76%] Building CXX object
CMakeFiles/rexx.dir/interpreter/platform/unix/TimeSupport.cpp.o
/Users/rvjansen/apps/oorexx/oorexx-code-0/interpreter/platform/unix/TimeSupport.cpp:62:5:
error: use of undeclared
identifier 'gettimeofday'
getti
I just checked in a new ErrorMessages.cpp file. See if that fixes the
problem.
Rick
On Sun, Sep 21, 2014 at 2:55 PM, René Jansen wrote:
> Generated with Cmake, now previous error gone, and I reach 70% of build.
> But then:
>
> */Users/rvjansen/apps/oorexx/oorexx-code-0/interpreter/platform/uni
Generated with Cmake, now previous error gone, and I reach 70% of build. But
then:
/Users/rvjansen/apps/oorexx/oorexx-code-0/interpreter/platform/unix/ErrorMessages.cpp:119:81:
error: use of
undeclared identifier 'CATD_ERR'
if ((catd = catopen(REXXMESSAGEFILE, SECOND_PARAMETER)
That's supposed to happen automatically when the list updates, but it might
be a good idea to start from a clean state.
Rick
On Sun, Sep 21, 2014 at 6:59 AM, René Jansen wrote:
> now you mention Cmake I might have to refresh my makefile from this
> cmake txt list.
> I'll try that later tod
now you mention Cmake I might have to refresh my makefile from this cmake
txt list.
I'll try that later today.
best regards,
René.
On 21 sep. 2014, at 00:43, Rick McGuire wrote:
> Well, you got me. There are no references to a StackClass in the
> CmakeList.txt file nor anywhere in the
Well, you got me. There are no references to a StackClass in the
CmakeList.txt file nor anywhere in the build tree. I have no idea where
this comes from.
Rick
On Sat, Sep 20, 2014 at 6:38 PM, René Jansen wrote:
> Now it comes a lot further, but it stops at:
>
> make: *** No rule to make targe
Now it comes a lot further, but it stops at:
make: *** No rule to make target `interpreter/classes/StackClass.cpp', needed
by `librexx_la-StackClass.lo'. Stop.
best regards,
René.
On 20 sep. 2014, at 22:48, Rick McGuire wrote:
> I also discovered that my fix will have the same problem bec
I also discovered that my fix will have the same problem because there is
also a PACKAGE_STRING define in the config.h. It looks like there is
something strange going on the the Mac build. On linux, the define is
#define PACKAGE ooRexx
Which didn't cause a syntax error. On your build, the ooRe
Never mind, I found it. Now I'm puzzled as to why Linux did not have the
same problem.
Rick
On Sat, Sep 20, 2014 at 4:41 PM, Rick McGuire wrote:
> These should be fixed now, but I'm curious about where the define for
> PACKAGE came from. I couldn't find it anywhere in the code. Is this in
>
These should be fixed now, but I'm curious about where the define for
PACKAGE came from. I couldn't find it anywhere in the code. Is this in
some generated config include?
Rick
On Sat, Sep 20, 2014 at 4:21 PM, René Jansen wrote:
> I have:
>
> In file included from ./interpreter/classes/ArrayC
I have:
In file included from ./interpreter/classes/ArrayClass.cpp:52:
In file included from ./interpreter/runtime/RexxCore.h:183:
In file included from ./interpreter/memory/RexxMemory.hpp:49:
./interpreter/memory/MemoryStack.hpp:204:5: warning: control reaches end of
non-void function [-Wreturn-
Ok, I'm able to get a clean build on Fedora. Please let me know if there
are any other build issues.
Rick
On Fri, Sep 19, 2014 at 6:46 PM, Rick McGuire wrote:
> I have to apologize, I was so excited about getting the base tests working
> on the 5.0 version that I merged the changes back in to
25 matches
Mail list logo
