EDIT

Also when rerunning on references I always get the same result (updated objects 2669), second time I would expect 0.

Can you check if there is still an error in the algorithm?

The same happens for taxon names (always 6 updated objects).

Please also check for taxa and specimens. I havn't tested yet.

Andreas Müller wrote:

Also when rerunning on references I always get the same result (updated objects 2669), second time I would expect 0.

Can you check if there is still an error in the algorithm?

The same happens for taxon names (always 6 updated objects).

Please also check for taxa and specimens. I havn't tested yet.

Maybe it is the same issue as for teams we found a few weeks ago. I will have a look.

Katja Luther wrote:

Andreas Müller wrote:

Also when rerunning on references I always get the same result (updated objects 2669), second time I would expect 0.

Can you check if there is still an error in the algorithm?

The same happens for taxon names (always 6 updated objects).

Please also check for taxa and specimens. I havn't tested yet.

Maybe it is the same issue as for teams we found a few weeks ago. I will have a look.

It is a problem of the cache update:

if (this.protectedAbbrevTitleCache == false){
           String oldAbbrevTitleCache = this.abbrevTitleCache;

           String newAbbrevTitleCache = cacheStrategy.getFullAbbrevTitleString(this);

           if ( oldAbbrevTitleCache == null   || ! oldAbbrevTitleCache.equals(newAbbrevTitleCache) ){
                this.setAbbrevTitleCache(null, false);
                String newCache = this.getAbbrevTitleCache();

                if (newCache == null){
                    logger.warn("New abbrevCache should never be null");
                }
                if (oldAbbrevTitleCache == null){
                    logger.info("oldAbbrevTitleCache was illegaly null and has been fixed");
                }
                result = true;
            }
        }
        return result;

first newAbbrevTitleCache is created by cacheStrategy.getFullAbbrevTitleString(this) and compared to the abbrevTitleCache of the reference and then the new cache is created by this.getAbbrevTitleCache(); and the strings are not always the same.

the problem for references is that if cacheStrategy.getFullAbbrevTitleString(this) returns null, the abbrevTitleCache is set to titleCache and then the comparison fails because the old abbrevTitleCache is the titleCache and the computed abbrevTitleCache is null.

Katja Luther wrote:

the problem for references is that if cacheStrategy.getFullAbbrevTitleString(this) returns null, the abbrevTitleCache is set to titleCache and then the comparison fails because the old abbrevTitleCache is the titleCache and the computed abbrevTitleCache is null.

Hm, in current rem_conf_am are 19 references with empty abbrevTitleCache but 40 objects (references) are updated according to the report.

I think we should further investigate here how to best handle this issue. However, this is another ticket. Can you create a new ticket for this? If the 2 above minor issues are fixed or moved to a "remaining" issue we can close this ticket.

Andreas Müller wrote:

Katja Luther wrote:

the problem for references is that if cacheStrategy.getFullAbbrevTitleString(this) returns null, the abbrevTitleCache is set to titleCache and then the comparison fails because the old abbrevTitleCache is the titleCache and the computed abbrevTitleCache is null.

Hm, in current rem_conf_am are 19 references with empty abbrevTitleCache but 40 objects (references) are updated according to the report.

I think we should further investigate here how to best handle this issue. However, this is another ticket. Can you create a new ticket for this? If the 2 above minor issues are fixed or moved to a "remaining" issue we can close this ticket.

The other references are references where the titleCache itself is null and during save/merge the titleCache is created by the CdmPreDataChangeListener

if (identifiableEntity.hasEmptyTitleCache()){
                    identifiableEntity.setTitleCache(identifiableEntity.toString(), false);
                }

Andreas Müller wrote:

If running on multiple classes the update result returns the total number of updated objects. Better we had them separated by classes, e.g. 10 taxon names updated, 5 authors/teams updated, ...

But this may go into a new ticket if not easy to fix.

This needs a separation of the different kind of objects because actually all updated objects are in one list. If we want to release soon, I would create a new ticket for this issue.(-> #8232)