Gregory Stark <stark@enterprisedb.com> writes:
> [ revised version of tuptoaster-fixup.patch ]
I've applied most of this. I didn't like the hardwired assumption about
minimum size to compress, and anyway you had put it in a place where it
broke the intended behavior for SET STORAGE PLAIN. I replaced it with
just a test to save palloc/pfree overhead in toast_compress_datum, which
has at least got knowledge of which PGLZ strategy is being used.
regards, tom lane