b2f0388 is the "biggie" ... Mark, this should fix your chroot issue.
I've tested this on all platforms in my QA farm with no PCP libraries
and no PCP headers installed.
Nathan, the perly bits may need some tweaking for mingw which is one
platform I cannot test at the moment.
Changes committed to git://oss.sgi.com/kenj/pcp.git dev
build/rpm/pcp.spec.in | 3
configure.in | 6
src/cpan/LogImport/Makefile.PL | 8
src/cpan/MMV/Makefile.PL | 8
src/cpan/PMDA/Makefile.PL | 8
src/include/.gitignore | 4
src/include/GNUmakefile | 27
src/include/builddefs.in | 3
src/include/impl.h | 1237 -----------------------------------
src/include/import.h | 67 -
src/include/mk_pmdbg | 62 -
src/include/mmv_dev.h | 84 --
src/include/mmv_stats.h | 107 ---
src/include/pcp/.gitignore | 4
src/include/pcp/GNUmakefile | 46 +
src/include/pcp/impl.h | 1237 +++++++++++++++++++++++++++++++++++
src/include/pcp/import.h | 67 +
src/include/pcp/mk_pmdbg | 62 +
src/include/pcp/mmv_dev.h | 84 ++
src/include/pcp/mmv_stats.h | 107 +++
src/include/pcp/platform_header.h.in | 611 +++++++++++++++++
src/include/pcp/platform_trailer.in | 14
src/include/pcp/pmafm.h | 51 +
src/include/pcp/pmapi.h | 728 ++++++++++++++++++++
src/include/pcp/pmda.h | 571 ++++++++++++++++
src/include/pcp/pmiestats.h | 41 +
src/include/pcp/pmimport.h | 52 +
src/include/pcp/pmtime.h | 111 +++
src/include/pcp/trace.h | 116 +++
src/include/pcp/trace_dev.h | 99 ++
src/include/pcp/trace_hash.h | 66 +
src/include/platform_header.h.in | 611 -----------------
src/include/platform_trailer.in | 14
src/include/pmafm.h | 51 -
src/include/pmapi.h | 728 --------------------
src/include/pmda.h | 571 ----------------
src/include/pmiestats.h | 41 -
src/include/pmimport.h | 51 -
src/include/pmtime.h | 111 ---
src/include/trace.h | 116 ---
src/include/trace_dev.h | 99 --
src/include/trace_hash.h | 66 -
src/pmdas/lustrecomm/libreadfiles.h | 6
src/pmdas/trace/GNUmakefile | 6
src/pmie/src/eval.c | 7
src/pmie/src/lexicon.c | 61 -
src/pmie/src/show.c | 2
47 files changed, 4140 insertions(+), 4092 deletions(-)
commit 72ce76ddabdf9e10475ded307c517aedde2034ff
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Feb 9 16:07:48 2012 +1100
RPM spec file - remove verbose diagnostics
Whatever problem this was debugging has either been fixed or has moved on.
commit b2f0388deffaaf4c8e9e98a4f3eecd76633e68f7
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Feb 9 10:47:49 2012 +1100
Rearrange headers in the source tree
These changes allow all of PCP to be built in an environment where
/usr/include/pcp does not exist and there are no PCP libraries in
/usr/lib* ... so everything PCP-related that used in the build comes
out of the PCP source tree.
Further, the references to the PCP headers are consistent throughout.
If the source file is included in the binary package (by way of an
example or demo), then the form
#include <pcp/pmapi.h>
is used to match the man pages. Otherwise (for the vast majority of
files) the form
#include "pmapi.h"
is used.
This style rule applies to all the PCP headers, not just pmapi.h.
With these changes, Mark's chroot build issue should be resolved.
Also in this commit, Perl build (Makefile.PL) changes to remove the
Note (probably harmless): No library found for ...
messages when PCP libraries are not installed in /usr/lib* ... these
messages would not stop the build, but DID produce bad Perl modules
with unresolved symbols ... I've seen the messages in some of the QA
environments (but probably only for the first build!), and now
understand what they mean!!
Nathan, a similar Perl build fix needs to be made for the Mingw builds
that I cannot test at the moment.
commit 2348bc272be5d2b2f80076925b112f67e68cda2a
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Feb 8 13:59:12 2012 +1100
pmie/lexicon.c - missing_break
Original Coverity case #79
http://people.redhat.com/mgoodwin/pcp-cov/1/79lexicon.c.html#error
There was no issue here, the control never passed to the "missing"
break was never executed. However cleaned up the code a little to
improve readability and replace some literal numeric constants by
local symbolic constants.
commit c45540fb3a8952721a3d94a73af9062d32293bfb
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sun Feb 5 07:14:59 2012 +1100
pmie/eval.c - missing_break
Original Coverity case #78
http://people.redhat.com/mgoodwin/pcp-cov/1/78eval.c.html#error
Cases in question (x->op == NULL and x->op == OP_VAR) can never
happen, so the missing break; was benign ... removed the dead code
and added a comment to explain.
commit d23043be5e90f5fa8344458a60eb62a19fd79f96
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sun Feb 5 06:23:27 2012 +1100
pmie/show.c - forward_null
Original Coverity case #64
http://people.redhat.com/mgoodwin/pcp-cov/1/64show.c.html#error
Code was correct, added an assert() to reinforce the fact that
x->arg1 is not NULL in the unary operator case.
|