Bug 544089

Created attachment 377888 [details]
Proposed patch to referint.c file

Created attachment 377890 [details]
The patched referint.c file

The referint.c file patched with the proposed patch.

https://bugzilla.redhat.com/attachment.cgi?id=377888&action=diff#referint.c_sec4

at line 392 in the new file - you have this:
		           (slapi_attr_value_find(attr, (struct berval *)slapi_value_get_berval(oldDNslv)) == 0))

is it really necessary to cast the return value of slapi_value_get_berval()?
int slapi_attr_value_find( const Slapi_Attr *a, const struct berval *v );
so it takes a const, and slapi_value_get_berval() returns a const.

No, you're right, it is not necessary. The thing is that the return type of slapi_value_get_berval() is not precisely indicated in RHDS plug-in guide (http://www.redhat.com/docs/manuals/dir-server/8.1/pdf/Plug-in_Guide.pdf). So i've searched an example in the server code. I've found it in ldap/servers/slapd/pw.c (lines 1000-10004) :
if (slapi_attr_value_find(attr, (struct berval *)slapi_value_get_berval(vals[0])) == 0 )

Now that i've looked into slapi-plugin.h the return type is indeed const struct berval *...

Thanks.  In the future, then plugin guide will be generated directly from slapi-plugin.h with doxygen.  There is some support already, but there is a lot of API to document . . .

Hi Andrey,

I'd like to work on bug 557224 "subtree rename breaks the referential integrity plug-in", which looks requiring to change the code this bug fix affects.

Could you please provide us a new patch including the change suggested by Rich?  We are going to check it.  That helps me a lot to work on the bug...

Thanks in advance,
--noriko

Created attachment 386694 [details]
Proposed patch to referint.c file

Modified patch including the change suggested by Rich

Comment on attachment 386694 [details]
Proposed patch to referint.c file

Thank you, Andrey!

Your new patch looks good to me.

Hi Noriko,

here is the new patch. You are right, it is much simpler to integrate it now. If we try to do that after the modifications necessary to take into account subtree renames it would be a nightmare...

Created attachment 386725 [details]
git patch file containing the fix provided by Andrey.

Thanks to Andrey for providing the patch.

Reviewed by rmeggins and nhosoi.

Pushed to master.

$ git merge work
Updating 0e9fd8b..1a76980
Fast forward
 ldap/servers/plugins/referint/referint.c |  152 ++++++++++++++++-------------
 1 files changed, 84 insertions(+), 68 deletions(-)
$ git push
Counting objects: 13, done.
Delta compression using 4 threads.
Compressing objects: 100% (7/7), done.
Writing objects: 100% (7/7), 1.86 KiB, done.
Total 7 (delta 4), reused 0 (delta 0)
To ssh://git.fedorahosted.org/git/389/ds.git
   0e9fd8b..1a76980  master -> master

verified - RHEL 4 

version:
redhat-ds-base-8.2.0-2010051204.el4dsrv

before modrdn:

# john.doe, People, example.com
dn: uid=john.doe,ou=People,dc=example,dc=com
uid: john.doe
sn: Doe
givenName: John
cn: DOE John
objectClass: inetorgperson
objectClass: organizationalPerson
objectClass: person
objectClass: top
manager: uid=small.chief,ou=People,dc=example,dc=com
manager;10: uid=big.chief,ou=People,dc=example,dc=com
manager;20: uid=small.chief,ou=People,dc=example,dc=com


after modrdn:

# john.doe, People, example.com
dn: uid=john.doe,ou=People,dc=example,dc=com
uid: john.doe
sn: Doe
givenName: John
cn: DOE John
objectClass: inetorgperson
objectClass: organizationalPerson
objectClass: person
objectClass: top
manager: uid=small.chief.jr,ou=people,dc=example,dc=com
manager;20: uid=small.chief.jr,ou=people,dc=example,dc=com
manager;10: uid=big.chief.jr,ou=people,dc=example,dc=com