Re: (ITS#7090) back-mdb produces wrong slapcat output

[hyc@symas.com]

michael@stroeder.com wrote:
> I've re-tested with git RE24 65a49595f563573f91da936ea27f5eaf6b21a48d with
> *nearly* the same data (data2) and the bug does not seem to happen again. Hmm...

You still haven't answered this first question:

 > If you revert the code to an older commit, does it produce the same result?

Since you haven't provided steps to reproduce the error, all we can do is rely 
on you to triage the problem.

One crucial step in isolating the problem would be to determine if the error 
actually occurs during slapadd or slapcat.

If you knew that the problem did not occur in a previous revision, then by 
slapadd'ing the LDIF with that known good revision, and then slapcat'ing with 
both the known revision and the later revision, would reveal whether the error 
is actually stored in the database, or whether it's a bug in how slapcat reads 
out the database.

> Also LDAP access to the entries mixed in slapcat output is completely ok.

This would imply that the error is isolated to slapcat. Since you claim the 
problem did not occur in an earlier revision, it might be easy to run thru the 
commit logs and find the cause of the problem. But without your confirmation 
of these guesses, we still have nothing to go on.

Since we have nothing to go on, and therefore could take no steps to address 
this issue, it should come as no surprise to you that testing an even later 
revision than the one you originally reported against still shows the bug 
occurring.

Bug-hunting takes systematic, methodical testing. You can't just go changing 
revisions willy-nilly.

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