MDMP 46C Combined UC/UPG, vocabulary assignment required pre-upgrade?

[boobboo]

I have a question about SPUM4 when doing an Combined UC&UPG on a 46c system.

As per the manual you have to run all the scans to complete SPUM4 in Prepare before starting the upgrade.

But do you have to perform the language assignment of the vocabulary list before the upgrade?

Is it possible to do this using SPUMG after the upgrade?

Basically I am being pushed for time and the project team is not ready to do the language assignments during Prepare. So it would be easier for me if I could do the scans and collate all the information required then perform the upgrade. Once the upgrade is complete, then allow the upgrade team to assign the languages post upgrade using the SPUMG transaction.

I am right in thinking that the vocabulary list has to be completely processed before conversion, this is because the words will use the global fallback which may not be right?

Thanks

Chris

[nils_buerckel]

Hi Chris,

you have to do the vocabulary assignment in the SPUM4 environment (on 4.6c), if you decide to use CU&UC. In CU&UC it does not make sense to do the assignment after the upgrade.

In case you go for TU&UC, you have to do the upgrade first and then do the vocabulary assignment.

Regarding vocabulary maintenance:

It would be possible to even leave ALL language fields empty in the vocab and go on with the scans. However this will probably produce millions of entries in the reprocess logs (after the reprocess scan), which cannot be handled. The same will be valid for transaction SUMG, which needs to be executed after the conversion (see UCG). This will result in a completely meaningless test for the SPUM4/SUMG.

If your scope for this test is 100% concentrated on export / import downtime, then this might make sense. However if you are looking for a complete CU&UC test, this does not make sense.

On the other hand side, there are only a few customers who really manage to maintain the vocab 100% - so a small percentage of unassigned words is usually left over. But the goal should be, to get this as low as possible !

Normally in the first trial, the number of the unassigned words in the vocab can be relatively high. But the result of the reprocess scan can be used to assign additional words in the vocab (and rerun the reprocess scan after additional maintenance).

Of course this will take time ! But if you want to avoid problems with many reprocess logs and repair logs, I would recommend to spend some time on maintenance of the vocab.

What happens, if a word is not assigned in the vocab ?

1) It will produce at least one (probably multiple) reprocess logs

2) R3load will use the fallback code page for the conversion of one or multple entries and it will write a message in the according xml-file

3) In SUMG, the according entries (again one or multiple) will show up in the manual repair

Best regards,

Nils Buerckel

Solution Management

Globalization Services

SAP AG

[boobboo]

Nils,

Thanks for that, we are discussing our options and will see how things go.

I am inclined to force the client to do as much on the vocabulary assignment as possible, purely because the fallback code page might lead to corruption on the return-trip reprocessing.

Thanks

Chris

[nils_buerckel]

Chris,

I would not call it "corruption", since all entries converted with the fallback code page can be repaired in any case in SUMG.

The problem will be the high amount of entries, if the vocabulary is not maintained properly ...

Regards,

Nils Buerckel

SAP AG