On 2013-01-07 18:27-0800 Alan W. Irwin wrote:
> After implementing the use of DESTDIR (as above) or the $ENV{DESTDIR}
> alternative in the install(CODE logic in doc/CMakeLists.txt, all the
> above issues went away. However, for some odd reason you do not get
> the desired result; even though the obvious errors go away, the
> install location is _still_ without the DESTDIR prefix.
It turned out I was conflating two issues which are probably separate.
My query to the cmake developers list got a quick and extremely useful
response from David Cole. The issue was I needed to escape the "$" in
$ENV{DESTDIR} so the build system would read the environment variable at make
time
when it is defined rather than at cmake time (where it was not defined
for this case).
Thus, revision 12284 follows David Cole's tip and
fixes the DESTDIR issue completely for the
non-parallel install case, i.e.,
make DESTDIR=/tmp/plplot_test install >& destdir.out
installs the documentation with the appropriate DESTDIR prefix and
without any other issues.
Furthermore, all is well
with the parallel install case without DESTDIR, e.g.,
make -j4 install >& install.out
works fine.
However, the error messages and segfault I referred to
in my previous message are still
there if you try a parallel install with DESTDIR, e.g.,
make DESTDIR=/tmp/plplot_test -j4 install >& destdir4.out
The segfault means we are likely dealing with some memory
management issue whose symptoms come and go. And another
source of uncertainty is parallel build issues often
depend critically on timing even if there is not a segfault
involved. So the "conclusions" below are probable ones
but not absolutely guaranteed.
This issue appears to have something to do with the cmake option
-DBUILD_DOX_DOC=ON which was used for all the above results.
If I turn that off (which is what happens
by default) the parallel install issue disappears. OTOH, I do not
encounter this issue with ephcom (where I have just fixed up how the
doxygen documentation is installed with DESTDIR in a similar way to
what I just did for PLplot). Note the ephcom build system concerning
doxygen documentation is simpler without some of the build system
machinery that occurs for PLplot in the -DBUILD_DOX_DOC=ON case. So
further debugging of the PLplot install with -DBUILD_DOX_DOC=ON and
for the DESTDIR case is indicated.
Unfortunately, I have run out of time for this (I must get back to
ephcom and te_gen release issues) so I hope someone else will take a
look further.
Alan
__________________________
Alan W. Irwin
Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).
Programming affiliations with the FreeEOS equation-of-state
implementation for stellar interiors (freeeos.sf.net); the Time
Ephemerides project (timeephem.sf.net); PLplot scientific plotting
software package (plplot.sf.net); the libLASi project
(unifont.org/lasi); the Loads of Linux Links project (loll.sf.net);
and the Linux Brochure Project (lbproject.sf.net).
__________________________
Linux-powered Science
__________________________
------------------------------------------------------------------------------
Master SQL Server Development, Administration, T-SQL, SSAS, SSIS, SSRS
and more. Get SQL Server skills now (including 2012) with LearnDevNow -
200+ hours of step-by-step video tutorials by Microsoft MVPs and experts.
SALE $99.99 this month only - learn more at:
http://p.sf.net/sfu/learnmore_122512
_______________________________________________
Plplot-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/plplot-devel
