Claudio,

>Abandoned for performance, but:
>- Deleting checksum space would mean an immediate ODS change, right?

Yes absolutely. But if you check the subject of this message it's
"Next ODS Change." My goal is to collect all the ideas we have for
ODS changes so they can be made at once. Certainly, saving 2 bytes
per page would not be a reason to change the ODS in and of itself.

>- Also, if someone wants to trade extra reliability for less performance,
>checksum can be enabled. If you remove those bytes, that tweak can't be done
>in the future. Make sure there's enough reasons to justify the gained space.

Right. But in this case, the performance hit is enormous and the
protection minimal. Perhaps we should increase the space so something
like the torn algorithm could be used?

>- What if someone wants the feature to track a ghost error that might come
>from faulty hardware or SMP operation or garbage collection bug, etc?

Garbage collection wouldn't show up, since the checksum (or whatever)
must be set long after the garbage has gone out. But sure, there might
be a case where turning on checksumming would be a reasonable diagnostic.

Regards,

Ann