Re: check_recovery_target_lsn() does a PG_CATCH without a throw

This has been committed.

On 2019-06-24 06:06, Michael Paquier wrote:
> I have been looking at this patch set. Patch 0001 looks good to me.
> You are removing all traces of a set of timestamp keywords not
> supported anymore, and no objections from my side for this cleanup.
>
> +#define MAXPG_LSNCOMPONENT 8
> +
> static bool
> check_recovery_target_lsn(char **newval, void **extra, GucSource source)
> Let's avoid the duplication for the declarations. I would suggest to
> move the definitions of MAXPG_LSNLEN and MAXPG_LSNCOMPONENT to
> pg_lsn.h. Funny part, I was actually in need of this definition a
> couple of days ago for a LSN string in a frontend tool. I would
> suggest renames at the same time:
> - PG_LSN_LEN
> - PG_LSN_COMPONENT

I ended up rewriting this by extracting the parsing code into
pg_lsn_in_internal() and having both pg_lsn_in() and
check_recovery_target_lsn() calling it. This mirrors similar
arrangements in float.c

--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services