Re: [HACKERS] generating fmgr prototypes automatically

On 1/3/17 2:16 PM, Pavel Stehule wrote:
> patch 0001 .. trivial cleaning
> patch 0002 .. renaming lo_* to be_lo_* -- the prefix "be" is not what I
> expect - maybe "pg" instead. More because the  be-fsstubs.h will be
> holds only lo_read, lo_write on end

It's the naming that the OpenSSL functions use, e.g., be_tls_init.

> patch 0003 .. trivial, but doesn't mean so we have not regress tests for
> these functions?

OK, added tests.

> patch 0004 .. bigger part - I miss a comments when there are a exceptions:
>
> extern Datum numeric_float8_no_overflow(PG_FUNCTION_ARGS);
> extern Datum nextval(PG_FUNCTION_ARGS);
> extern Datum fmgr_sql(PG_FUNCTION_ARGS);
> extern Datum aggregate_dummy(PG_FUNCTION_ARGS);

These functions and their special purpose are commented in the .c files,
so I think that is covered OK.  I added an additional comment to the
numeric_float8_no_overflow().

> I found some obsolete definitions in c files

OK, fixed those as well.  I think I initially only looked in .h files.
That just underlines how inconsistent this is, e.g.,

> pgstatfuncs.c
>
> /* bogus ... these externs should be in a header file */

> namespace.c
>
> /* These don't really need to appear in any header file */

New patch set attached.

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