Validation on upgradeprovision patches needed

[Matthieu Patou]

  On 11/07/2010 22:38, Matthias Dieter Wallnöfer wrote:
> Hi ekacnet,
>
> my comments:
>
> (repl_meta_data.c)
>> static uint64_t find_max_local_usn(struct replPropertyMetaDataBlob omd)
>> {
>>        int count = omd.ctr.ctr1.count;
>>        uint64_t max = 0;
>>        int i;
>>        for (i=0; i < count; i++) {
>>                struct replPropertyMetaData1 m = omd.ctr.ctr1.array[i];
>>                if (max < m.local_usn) {
>>                        max = m.local_usn;
>>                }
>>        }
>>        return max;
>> }
> ^^^ "count" and "i" should be from type "uint32_t" (these are DSDB 
> counters).
I guess it could be
>
> (objectclass_attrs.c)
>>                         found = str_list_check(may_contain, 
>> attr->lDAPDisplayName);
>>                 }
>>                 if (!found) {
>> +                       found = str_list_check(harmless_attrs, 
>> attr->lDAPDisplayName);
>> +               }
>> +               if (!found) {
>>                         ldb_asprintf_errstring(ldb, 
>> "objectclass_attrs: attribute '%s' on entry '%s' does not exist in 
>> the specified objectclasses!",
>>                                                msg->elements[i].name,
>>                                                
>> ldb_dn_get_linearized(msg->dn));
> ^^^ Is this really the best solution? Couldn't "upgradeprovision" 
> delete all "parentGUIDs" to be consistent with recent provisions?
>
This in fact is tridge proposal that I just picked (if was a simple diff 
with no commit message), globaly his position is that we have already 
the cleanning code in one dsdb module so there shouldn't be a need for a 
cleaning done by upgradeprovision.

After it's up to tridge and co. to decide the cleanest way to do ...