On 07/23/10 11:59 AM, François Bissey wrote:
I just tried to build Sage, and got a failure with building IML. See log
here.
http://boxen.math.washington.edu/home/kirkby/iml-1.0.1.p12.log
As soon as I restarted "make" again, so the build completed ok.
On thing I did do, was reverse the order of the list of items listed under
'all' in deps. i.e.
all: $(BASE) \
$(INST)/$(ZODB) \
$(INST)/$(ZNPOLY) \
$(INST)/$(ZLIB) \
$(INST)/$(WEAVE) \
$(INST)/$(TWISTED) \
$(INST)/$(TERMCAP) \
$(INST)/$(TACHYON) \
$(INST)/$(SYMPY) \
...etc etc
$(INST)/$(ATLAS)
instead of the previous
all: $(BASE) \
$(INST)/$(ATLAS) \
$(INST)/$(BLAS) \
$(INST)/$(BOEHM_GC) \
$(INST)/$(BOOST_CROPPED) \
$(INST)/$(CDDLIB) \
$(INST)/$(CEPHES) \
$(INST)/$(CLIQUER) \
... etc etc
$(INST)/$(ZODB)
This was an attempt to change the build order, with a hope of finding
problems with 'deps'. However, that log does not look like a failure
because of a dependency missing. In fact, I've no idea what the error is.
But I do know it cleared up as soon as I typed "make" again.
The system is a Sun Ultra 27, 3.33 GHz quad core, 8 threads. OpenSolaris
06/2009. 1000 (yes) 1000 threads were used.
This was a parallel build of packages.
Looks like a parallel make issue, which is why it worked when you restarted
make.
Some of the objects used probably weren't ready yet and their build completed
after the building of these 2 particular objects died.
So when you restarted it was fine. Cannot really inspect that in depth right
now.
Francois
If that is so, I think we should unset MAKE in IML, if it can't reliably be
built in parallel.
Any thoughts from anyone else?
Dave
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
