On Tue, Sep 17, 2019 at 11:29:13AM +0900, Michael Paquier wrote:
> The code paths of the patch calling pg_strtouint64_check to parse
> completion tags (spi.c and pg_stat_statements.c) should complain in
> those cases as the format of the tags for SELECT and COPY is fixed.
>
> In order to unify the parsing interface and put all the conversion
> routines in a single place, I still think that the patch has value so
> I would still keep it (with a fix for the queryId parsing of course),
> but there is much more to it.
Forgot to mention that another angle of attack would be of course to
add some control flags in the optimized parsing functions to make them
more permissive regarding the handling of the trailing characters, by
not considering as a syntax error the case where the last character is
not a zero-termination.
--
Michael
