[GRASS-dev] make on SMP [was: GRASS 6.3.0 release preparation]

[Soeren Gebbert]

Hi,
i compiled the latest grass cvs version today several times on a dual
core intel machine
with "make -j4". The home directory is located on a separate server
mounted via nfs.
Three builds with no errors, one build with an error in r.terraflow.
Make output:

...
GRASS GIS compilation log
-------------------------
Started compilation: Wed Oct 24 17:47:03 CEST 2007
--
Errors in:
/homes/excoll/soeren/src/grass_test_build/grass6/raster/r.terraflow
--
In case of errors please change into the directory with error and run 'make'.
...

At line 6700 + something the following errors occur:
...
OBJ.i686-pc-linux-gnu/FLOAT/sweep.o: In function `AMI_STREAM':
/homes/excoll/soeren/src/grass_test_build/grass6/raster/r.terraflow/./IOStream/include/ami_stream.h:220:
undefined reference to `ami_single_temp_name(char*, char*)'
/homes/excoll/soeren/src/grass_test_build/grass6/raster/r.terraflow/./IOStream/include/ami_stream.h:221:
undefined reference to `open_stream(int, AMI_stream_type)'
/homes/excoll/soeren/src/grass_test_build/grass6/raster/r.terraflow/./IOStream/include/ami_stream.h:256:
undefined reference to `open_stream(char*, AMI_stream_type)'
/homes/excoll/soeren/src/grass_test_build/grass6/raster/r.terraflow/./IOStream/include/ami_stream.h:220:
undefined reference to `ami_single_temp_name(char*, char*)'
/homes/excoll/soeren/src/grass_test_build/grass6/raster/r.terraflow/./IOStream/include/ami_stream.h:221:
undefined reference to `open_stream(int, AMI_stream_type)'
collect2: ld returned 1 exit status
make[2]: *** [/homes/excoll/soeren/src/grass_test_build/grass6/dist.i686-pc-linux-gnu/bin/r.terraflow]
Error 1
...

The full log is located here:
http://www-pool.math.tu-berlin.de/~soeren/grass/files/grass_build_24_oct_2007_17_46.log

Soeren

2007/10/21, Glynn Clements <glynn at gclements.plus.com>:
>
> Maciej Sieczka wrote:
>
> > > But that means rewriting every single Makefile.
> >
> > Glynn
> >
> > After your recent commits in CVS I'm able to build GRASS 6.3
> > with make -j3 without any errors, no matter that one core
> > and HDD are pretty busy with other things. Does that mean
> > you managed to fix this issue or was I just lucky?
>
> I think that the most recent change to Module.make will have fixed a
> lot of the problems.
>
> OTOH, I've just fixed[1] some problems with NVIZ' Makefiles which
> could have caused problems on parallel builds, and it's just luck that
> you avoided those.
>
> There may still be problems lurking. When it comes to parallel make,
> testing is a fairly hit-and-miss concept, as most bugs depend upon
> exactly when particular commands get executed, and it's practically
> impossible to generate specific test cases.
>
> [1] Or maybe made them worse. The reason I left NVIZ alone until now
> was that the Makefiles were a mess. Although I fixed some obvious
> problems, I may have introduced some new ones while cleaning up.
>
> --
> Glynn Clements <glynn at gclements.plus.com>
>
> _______________________________________________
> grass-dev mailing list
> grass-dev at grass.itc.it
> http://grass.itc.it/mailman/listinfo/grass-dev
>