Jacob Champion <pchampion@pivotal.io> writes:
> Patch looks good to me. Seems like a useful feature, and I agree that the two-pass implementation makes the change
veryeasy to review.
LGTM too. I made a couple of cosmetic improvements and pushed it.
> Quick note on test coverage: gcov marks the "needle not found" branch (the one marked `/* special case if fldsep not
foundat all */`) as being completely uncovered. I don't think that needs to gate this patch; it looks like it was
uncoveredbefore this feature was added.
We seem to be trying for full test coverage of this function now,
so I added a test case for that branch too.
> Doc builds are currently failing due to what appears to be an xmllint failure:
Unrelated, but see
https://www.postgresql.org/message-id/flat/E2EE6B76-2D96-408A-B961-CAE47D1A86F0%40yesql.se
regards, tom lane