public inbox for [email protected]  

help / color / mirror / Atom feed

libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()?
4+ messages / 2 participants
[nested] [flat]

* libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()?
@ 2024-12-21 12:54 Jeroen Vermeulen <[email protected]>
  2024-12-21 15:56 ` Re: libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Tom Lane <[email protected]>
  0 siblings, 1 reply; 4+ messages in thread

From: Jeroen Vermeulen @ 2024-12-21 12:54 UTC (permalink / raw)
  To: pgsql-interfaces

Any objections to changing these two libpq functions' PGresult* parameters
to be const?

So "char * PQcmdStatus(PGresult *)" would become "char *PQcmdStatus(const
PGresult *)", and "char *PQcmdTuples(PGresult *)" would become "char
*PQcmdTuples(const PGresult *)".


Jeroen


^ permalink  raw  reply  [nested|flat] 4+ messages in thread

* Re: libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()?
  2024-12-21 12:54 libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Jeroen Vermeulen <[email protected]>
@ 2024-12-21 15:56 ` Tom Lane <[email protected]>
  2024-12-21 15:59   ` Re: libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Jeroen Vermeulen <[email protected]>
  2024-12-21 16:06   ` Re: libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Tom Lane <[email protected]>
  0 siblings, 2 replies; 4+ messages in thread

From: Tom Lane @ 2024-12-21 15:56 UTC (permalink / raw)
  To: Jeroen Vermeulen <[email protected]>; +Cc: pgsql-interfaces

Jeroen Vermeulen <[email protected]> writes:
> Any objections to changing these two libpq functions' PGresult* parameters
> to be const?

> So "char * PQcmdStatus(PGresult *)" would become "char *PQcmdStatus(const
> PGresult *)", and "char *PQcmdTuples(PGresult *)" would become "char
> *PQcmdTuples(const PGresult *)".

Both of those return pointers to res->cmdStatus, so it would not make
any sense to const-ify the argument unless you also const-ify the
result; which would cause compiler warnings for users.  I'm pretty
sure we've explicitly rejected making this change for that reason.

			regards, tom lane





^ permalink  raw  reply  [nested|flat] 4+ messages in thread

* Re: libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()?
  2024-12-21 12:54 libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Jeroen Vermeulen <[email protected]>
  2024-12-21 15:56 ` Re: libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Tom Lane <[email protected]>
@ 2024-12-21 15:59   ` Jeroen Vermeulen <[email protected]>
  1 sibling, 0 replies; 4+ messages in thread

From: Jeroen Vermeulen @ 2024-12-21 15:59 UTC (permalink / raw)
  To: Tom Lane <[email protected]>; +Cc: pgsql-interfaces

Ah, that explains.  Thanks.

On Sat, Dec 21, 2024, 16:56 Tom Lane <[email protected]> wrote:

> Jeroen Vermeulen <[email protected]> writes:
> > Any objections to changing these two libpq functions' PGresult*
> parameters
> > to be const?
>
> > So "char * PQcmdStatus(PGresult *)" would become "char *PQcmdStatus(const
> > PGresult *)", and "char *PQcmdTuples(PGresult *)" would become "char
> > *PQcmdTuples(const PGresult *)".
>
> Both of those return pointers to res->cmdStatus, so it would not make
> any sense to const-ify the argument unless you also const-ify the
> result; which would cause compiler warnings for users.  I'm pretty
> sure we've explicitly rejected making this change for that reason.
>
>                         regards, tom lane
>


^ permalink  raw  reply  [nested|flat] 4+ messages in thread

* Re: libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()?
  2024-12-21 12:54 libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Jeroen Vermeulen <[email protected]>
  2024-12-21 15:56 ` Re: libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Tom Lane <[email protected]>
@ 2024-12-21 16:06   ` Tom Lane <[email protected]>
  1 sibling, 0 replies; 4+ messages in thread

From: Tom Lane @ 2024-12-21 16:06 UTC (permalink / raw)
  To: Jeroen Vermeulen <[email protected]>; +Cc: pgsql-interfaces

I wrote:
> Both of those return pointers to res->cmdStatus, so it would not make
> any sense to const-ify the argument unless you also const-ify the
> result; which would cause compiler warnings for users.  I'm pretty
> sure we've explicitly rejected making this change for that reason.

[ digs in archives... ]

See previous discussions here:

https://www.postgresql.org/message-id/flat/NCBBLFLGLEIDHCJLONHNKEDGFJAA.mdklatt%40ou.edu

and here:

https://www.postgresql.org/message-id/flat/E1RaRtZ-0008aS-MK%40hans3

			regards, tom lane






^ permalink  raw  reply  [nested|flat] 4+ messages in thread

end of thread, other threads:[~2024-12-21 16:06 UTC | newest]

Thread overview: 4+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2024-12-21 12:54 libpq: make PGresult* "const" in PQcmdStatus()/PQcmdTuples()? Jeroen Vermeulen <[email protected]>
2024-12-21 15:56 ` Tom Lane <[email protected]>
2024-12-21 15:59   ` Jeroen Vermeulen <[email protected]>
2024-12-21 16:06   ` Tom Lane <[email protected]>

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox