I've got table buffering enabled, and multilocks on. I read the Help file and Tableupdate() says that it leaves the record pointer on the last record updated (presumably internally) which is necessarily the record you were just editing. The solution was to alter my base class ".save()" routine to reset the record pointer after doing any necessary commits. Similarly for Tablerevert()
But thanks for input Vlad.

BTW, that's a GREAT hairstyle to go with "Vlad" for a name. :-)

>Hi!
>
>It is also possible that you have a trigger for updating in the database and code in that trigger moves record pointer.
>
>>OK, I just read the help, and yes, tableupdate() moves the record pointer. Sheesh, I wouldn't have expected that.
>>Oh, well.
>>
>>>I have an instance where the line
>>>

>>>tableUpdate(m.allrows, .t., m.cAlias)
>>>

>>>is moving the record pointer in m.cAlias, eg might be on record 4 before the call but end up on record 8 afterward. Sure, changes have occurred to more than one record, but should this command leave the record point alone??
>>>Is this to be expected??