Re: WAL, xl_heap_insert and tuple oid mystry - Mailing list pgsql-hackers

(2011/04/13 8:34), Tom Lane wrote:
> Greg Stark<gsstark@mit.edu>  writes:
>> Doesn't pg_lesslog depend on this?
> 
> One hopes not.

AFAIK it's safe because pg_lesslog removes full-page image only when it
has enough information for substitute incremental log.

For example of XLOG_HEAP_INSERT, pg_lesslog generates incremental logs
such as xl_heap_header and tuple image from the inserted tuple in the
full-page image, and replaces full-page image with them.

Regards,
-- 
Shigeru Hanada