On 23.10.2018 0:53, Heikki Linnakangas wrote:
> I'd expect the decompression to read from the on-disk buffer, and unpack
> to readRecordBuf, I still don't see a need to copy the packed record to
> readRecordBuf. If there is a need for that, though, the patch that
> implements the packing or compression can add the memcpy() where it
> needs it.
I agree with it. Eventually, placement of the WAL-record can be defined
by comparison the record, readBuf and readRecordBuf pointers.
In attachment new version of the patch.
--
Andrey Lepikhov
Postgres Professional
https://postgrespro.com
The Russian Postgres Company