Looks pretty good. You seem to have added a blank line at the end of postgres_fdw.c, which should be stripped back out.
Done.
> I'm not too keen on having *no* new regression tests, but defer to your > judgement.
So... I'm not *opposed* to regression tests. I just don't see a reasonable way to write one. If you've got an idea, I'm all ears.
The possible tests might be:
- creating a server/table with fetch_size set
- altering an existing server/table to have fetch_size set
- verifying that the value exists in pg_foreign_server.srvoptions and pg_foreign_table.ftoptions.
- attempting to set fetch_size to a non-integer value
None of which are very interesting, and none of which actually test what fetch_size was actually used.
But I'm happy to add any of those that seem worthwhile.
> Still not sure what you mean by remote execution options. But it might be > simpler now that the patch is closer to your expectations.
I'm talking about the documentation portion of the patch, which regrettably seems to have disappeared between v2 and v3.
Ah, didn't realize you were speaking about the documentation, and since that section was new, I wasn't familiar with the name. Moved.
...and not sure why the doc change didn't make it into the last patch, but it's in this one.
I also moved the paragraph starting "When using the extensions option, it is the user's responsibility that..." such that it is in the same varlistentry as "extensions", though maybe that change should be delegated to the patch that created the extensions option.