Обсуждение: Re: Fix misuse use of pg_b64_encode function (contrib/postgres_fdw/connection.c)

On 16.01.25 02:12, Ranier Vilela wrote:
> Per Coverity.
> 
> CID 1590024:    (CHECKED_RETURN)
> Calling "pg_b64_encode" without checking return value (as is done 
> elsewhere 8 out of 10 times).
> 
> The function *pg_b64_encode* has in the comments:
> [0]  "and -1 in the event of an error"
> 
> So, the function can fail.
> All other calls check the return, In this case it could not be different.
> 
> Fix by checking the return and reporting a message to the user,
> in case of failure.

Thanks, fixed.  (I changed the ereports to elogs, which is how other 
call sites do it.)

I also fixed a related problem in the pg_b64_decode() calls in libpq.

Maybe we could put a pg_nodiscard attribute on pg_b64_encode() and 
pg_b64_decode()?

> [0] I think the most correct would be *or* not *and* word?

I think both are ok here.

On 16.01.25 11:23, Ranier Vilela wrote:
> 
> 
> Em qui., 16 de jan. de 2025 às 05:07, Peter Eisentraut 
> <peter@eisentraut.org <mailto:peter@eisentraut.org>> escreveu:
> 
>     On 16.01.25 02:12, Ranier Vilela wrote:
>      > Per Coverity.
>      >
>      > CID 1590024:    (CHECKED_RETURN)
>      > Calling "pg_b64_encode" without checking return value (as is done
>      > elsewhere 8 out of 10 times).
>      >
>      > The function *pg_b64_encode* has in the comments:
>      > [0]  "and -1 in the event of an error"
>      >
>      > So, the function can fail.
>      > All other calls check the return, In this case it could not be
>     different.
>      >
>      > Fix by checking the return and reporting a message to the user,
>      > in case of failure.
> 
>     Thanks, fixed.  (I changed the ereports to elogs, which is how other
>     call sites do it.)
> 
> Thank you.
> 
> 
>     I also fixed a related problem in the pg_b64_decode() calls in libpq.
> 
>     Maybe we could put a pg_nodiscard attribute on pg_b64_encode() and
>     pg_b64_decode()?
> 
> +1

done