Re: Allow file inclusion in pg_hba and pg_ident files

Hi,

On Sat, Feb 26, 2022 at 03:04:43PM +0900, Michael Paquier wrote:
> On Wed, Feb 23, 2022 at 09:44:58AM -0800, Nathan Bossart wrote:
> > On Wed, Feb 23, 2022 at 12:59:59PM +0800, Julien Rouhaud wrote:
> >> 0001 adds a new pg_ident_file_mappings view, which is basically the same as
> >> pg_hba_file_rules view but for mappings.  It's probably already useful, for
> >> instance if you need to tweak some regexp.
> > 
> > This seems reasonable.
> 
> Interesting.  One can note that hba.c is already large, and this makes
> the file larger.  I'd like to think that it would be better to move
> all the code related to the SQL functions for pg_hba.conf and such to
> a new hbafuncs.c under adt/.  Would that make sense?

I'm fine with it.  Assuming that you meant to move also the underlying
functions that goes with it (fill_hba_line and such), that would end up
removing about 15% of hba.c (after applying 0001, 0002 and 0003).

Note that in order to do so we would need to expose quite a lot more about hba
internals, like tokenize_file() and parse_hba_line(), along with structs
HbaToken and TokenizedLine.