On Fri, Dec 25, 2020 at 09:04:41AM +0900, Michael Paquier wrote:
> Looks like the defense put in place by 6b1c5ca has allowed to catch up
> a bug here. When base64 has been copied from encode.c to src/common/
> for SCRAM (newlines should not be handled by SCRAM, hence the copy),
> we have done the same. The copied code just returns -1 for error
> paths. For this case, I think that you should also prefix those
> functions with "pg_", and also include the encode part for
> completeness.
I now understand the wisdom of your suggestion. Attached is a patch
that removes hex_decode from ecpg properly, and returns -1 from the
/common version.
--
Bruce Momjian <bruce@momjian.us> https://momjian.us
EnterpriseDB https://enterprisedb.com
The usefulness of a cup is in its emptiness, Bruce Lee