> That's wrong! What is normally does is read the last committed version of
> that
> record. Writers don't block readers.

hmm... don't know, but think he is speaking about uncommitted version.

On documentation we can find this:

isc_tpb_rec_version Enables an isc_tpb_read_committed transaction to read the most recently
committed version of a record even if other, uncommitted versions are
pending.

isc_tpb_no_rec_version Enables an isc_tpb_read_committed transaction to read only the latest
committed version of a record. If an uncommitted version of a record is
pending and isc_tpb_wait is also specified, then the transaction waits for
the pending record to be committed or rolled back before proceeding.
Otherwise, a lock conflict error is reported at once.

by comments of Cristoph = "the record with id 10104 isn't already committed".
Like Dimitry said, no_rec_version throws a error is normal. But if I understand what Cristoph said, Rec_version too is throwing errors. Note that by his comments Dimitry said this will not show error. This is strange, no?

I bet that something is commiting his first transaction, or transactions is not really record_version.

> I don't get your description of:
> I just want the 2nd transaction to "overread" pending changes from others

I don't get this too.. but

>
> What does that mean? Do you want to read the data that's available or do you
> want that reader to read the current uncommitted data? (dirty read)

Think he want that read only the data that IS committed and "the record with id 10104 isn't already committed".

>
> With regards,
>
> Martijn Tonies
> Upscene Productions
> http://www.upscene.com
>
> Download Database Workbench for Oracle, MS SQL Server, Sybase SQL
> Anywhere, MySQL, InterBase, NexusDB and Firebird!
>
> Database questions? Check the forum:
> http://www.databasedevelopmentforum.com
>