Re: ArrayIndexOutOfBoundsException in ContentField.activate(...)

Hi Serge!

Finally I could reproduce the exception I was mentioning a month ago. It 
somehow happens that for some fields there are two same entries in the 
database. Here is an excerpt of our JAHIA_FIELDS_DATA table.

ID_JAHIA_FIELDS_DATA
JAHIAID_JAHIA_FIELDS_DATA
PAGEID_JAHIA_FIELDS_DATA
CTNID_JAHIA_FIELDS_DATA
FIELDDEFID_JAHIA_FIELDS_DATA
TYPE_JAHIA_FIELDS_DATA
CONNECTTYPE_JAHIA_FIELDS_DATA
VALUE_JAHIA_FIELDS_DATA
RANK_JAHIA_FIELDS_DATA
RIGHTS_JAHIA_FIELDS_DATA
VERSION_ID
WORKFLOW_STATE
LANGUAGE_CODE
10263
1
1
10043
10041
1
0
100
0
11685
0
2
shared
10263
1
1
10043
10041
1
0
100
0
11685
0
2
shared
10264
1
1
10043
10042
20
0

0
11686
0
2
shared
10264
1
1
10043
10042
20
0

0
11686
0
2
shared
10265
1
1
10043
10043
20
0

0
11687
0
2
shared
10265
1
1
10043
10043
20
0

0
11687
0
2
shared
10266
1
1
10043
10039
2
0
test
0
11688
0
2
de
10267
1
1
10043
10040
20
0

0
11689
0
2
shared

For those fields, which have two entries, we get the mentioned exceptions. 
How come, that there are two same entries? I guess it is not a normal 
behaviour and how can we prevent this exception, which blocks the workflow 
as a page cannot be activated anymore?

Greetings,
Benjamin

Serge wrote on Fri, 13 May 2005 02:17:07 -0700

Hi Benjamin,

Do you have a reliable way of reproducing this so that we may test it ? We 
haven't be able to reproduce it here during our testing. Also which JVM 
are you using ?

Regards,
 Serge Huber.

Benjamin wrote on Wed, 11 May 2005 03:09:32 -0700 :

We are still using Jahia 4.0.5, but on heavy testing we encounter that the 
workflow process is quite unstable sometimes. When looking in the LOG 
files, I for instance can see an ArrayIndexOutOfBoundsException in the 
ContentField.acivate(...) method.

In this loop we get this exception in the line 
activeAndStagingEntryStates.remove (i);, because i has the value -1. It 
seem strange anyway that the removing from the activeAndStagingEntryStates 
collection is happening in the inner loop and not in the outer loop. As 
you haven't changed the code in the 4.1 version it seems to work for 
everybody else, or could it be that there is a bug?

// we backup all active entries that will be replaced
for (int i = 0; i < activeAndStagingEntryStates.size (); i++) {
ContentObjectEntryState actEntryState = (ContentObjectEntryState) 
activeAndStagingEntryStates.get (
i);
if (actEntryState.isActive ()) {
for (int j = 0; j < stagedEntries.size (); j++) {
ContentObjectEntryState staEntryState = (ContentObjectEntryState) 
stagedEntries.get (
j);
if (actEntryState.getLanguageCode ().equals (
staEntryState.getLanguageCode ())) {
// ok appollo, we catched an active entry here! we handle this..
if (versioningEnabled) {

ContentObjectEntryState backupEntryState = ContentFieldDB.getInstance 
().backupDBValue (
this, actEntryState); // we backup the active version
// ok we backuped an entry of the field
if (backupEntryState != null) {
// if the entry state list is loaded, let's update it
if (versioningEntryStates != null) {
versioningEntryStates.add (backupEntryState);
}
// also we must call the changeEntryState because, like, we changed the 
version or
// something hehe..
this.changeEntryState (actEntryState, backupEntryState,
jParams, stateModifContext);
}
} else {
// if versioning is not enabled, we delete the old value content!
this.deleteEntry (actEntryState);
}
// we then delete the active entry
ContentFieldDB.getInstance ().deleteDBValue (this, actEntryState);
activeAndStagingEntryStates.remove (i);
i -= 1;
}
}
}
}

Greetings,
Benjamin Papez