Re: test006-acl fails in HEAD

[Pierangelo Masarati <ando@sys-net.it>]

Armbrust, Daniel C. wrote:

> The system is a Fedora Core 2 box (actually just upgraded from Redhat 9 yesterday), on dual cpu amd.
>
> The only option I passed in with configure was an install location, otherwise everything else was left at the default.  I'm using 4.2.52 of BerkleyDB (unpatched - by the way - because at the time I built it the patches didn't look relevant.. But recent conversation on the mailing list shows that are relevant, especially for doing many multithreaded writes... Its on my todo list to fix that - but I'm nervous that I will have to reload all my databases if I rebuild Berkeley to have everything consistent, and I don't have time to do it right now)
>  
This should really have anything to do.  I vaguely recall that in
the past there was some dependency between ACL sets and ACIs,
and I always build with --enable-aci even if I haven't been using
them for a while.  However, at a first glance, now the sets code
does not require --enable-aci to be compiled straight in.

> However, I have never had the make tests fail on older versions of openldap, and they were using the same unpatched Berkeley.
>
> I ran "make test -d 128" (is this the correct way to change the debug level for the test?)
>  
I meant slapd's output, not the test script's output, sorry; I should
have been more clear.  Run (with bash; note 384 = 128+256, acl + stats)

export SLAPD_DEBUG=384
./run test006

from the tests/ directory and post the file testrun/slapd.1.log

Thanks, p.




   SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497