Turned out there was an LD_LIBRARY_PATH used for the already running openldap, but was set, so the "make test" prior to installation, was run against the installed libraries, instead of the newly compiled (but as yet uninstalled) ones in the build directory.

Arguably the test could "unset LD_LIBRARY_PATH" but that variable might be used to find berkeley dn or other thrid part libraries, so maybe LD_LIBRARY_PATH="`pwd`/libraries/liblber/.libs `pwd`/libraries/libldap/.libs" to the script that launches tests ?

Has there ever been any thought of having a --with-bdb=<prefix> which implies the addition of CPPFLAGS="-I<prefix>/include" and LDFLAGS="-L<prefix>/lib" and similarly for things like --with-ssl=<another prefix> ?

Personally i alway have to add -R<prefix>/lib also, but this varies by platform, eg. linux has removed this in preference of a gcc option which passes options through to the linker, so it's probably hard to do in a platform-agnostic manner.

On Fri, Mar 9, 2012 at 12:07 PM, Howard Chu hyc@symas.com wrote:

Quanah Gibson-Mount wrote:

...

--On Friday, March 09, 2012 10:49 AM +1000 "Brett @Google" brett.maxfield@gmail.com wrote:

Hello,

...

Are the tests in test050 (multimaster concurrency) for release 2.4.30, still likely or expected to fail on solaris 10 with bdb 4.8.30 ?

We dont use multimaster, but at one point in recent history the tests in test050 (multimaster concurrency) were expected to fail.

I'll log an ITS if it is now unexpected for this test to fail..

Cheers Brett

Waiting 5 seconds for slapd to start... Waiting 5 seconds for slapd to start... 29490 Segmentation Fault - core dumped

segmentation faults are never expected. You should file an ITS with the backtrace on the core file.

And the last few lines of slapd.1.log.

-- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/**project/http://www.openldap.org/project/