Re: Refactor query normalization into core query jumbling

> === 1
>
> + SetConstantLengths((JumbleState *) jstate, query, query_loc);
>
> This cast seems unnecessary.

Right. This is unnecessary. I will remove in the next iteration.

> === 2
>
> +CompLocation(const void *a, const void *b)
>
> In the commit message I can see "Functions are renamed to match core naming
> conventions" but wasn't comp_location() better?

Not sure if it's better or worse. I aimed for consistency here and
used pascal case to
match.

> === 3
>
> + /*
> + * generate the normalized query. Note that the normalized
> + * representation may well vary depending on just which
> + * "equivalent" query is used to create the hashtable entry. We
> + * assume this is OK.
> + */
> + norm_query = GenerateNormalizedQuery(jstate, query,
>
> Should part of this comment be on top of the GenerateNormalizedQuery()
> definition instead?

No, because this talks about the hashtable entry which is specific to
pg_stat_statements.

--
Sami Imseih
Amazon Web Services (AWS)