From: | Peter Eisentraut <peter(at)eisentraut(dot)org> |
---|---|
To: | pgsql-committers(at)lists(dot)postgresql(dot)org |
Subject: | pgsql: Improve pgbench definition of yyscan_t |
Date: | 2025-09-12 06:22:04 |
Message-ID: | E1uwxAa-000Ik3-09@gemulon.postgresql.org |
Views: | Whole Thread | Raw Message | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-committers |
Improve pgbench definition of yyscan_t
It was defining yyscan_t as a macro while the rest of the code uses a
typedef with #ifdef guards around it. The latter is also what the
flex generated code uses. So it seems best to make it look like those
other places for consistency.
The old way also had a potential for conflict if some code included
multiple headers providing yyscan_t. exprscan.l includes
#include "fe_utils/psqlscan_int.h"
#include "pgbench.h"
and fe_utils/psqlscan_int.h contains
#ifndef YY_TYPEDEF_YY_SCANNER_T
#define YY_TYPEDEF_YY_SCANNER_T
typedef void *yyscan_t;
#endif
which was then followed by pgbench.h
#define yyscan_t void *
and then the generated code in exprscan.c
#ifndef YY_TYPEDEF_YY_SCANNER_T
#define YY_TYPEDEF_YY_SCANNER_T
typedef void* yyscan_t;
#endif
This works, but if the #ifdef guard in psqlscan_int.h is removed, this
fails.
We want to move toward allowing repeat typedefs, per C11, but for that
we need to make sure they are all the same.
Reviewed-by: Chao Li <li(dot)evan(dot)chao(at)gmail(dot)com>
Discussion: https://www.postgresql.org/message-id/flat/10d32190-f31b-40a5-b177-11db55597355(at)eisentraut(dot)org
Branch
------
master
Details
-------
https://git.postgresql.org/pg/commitdiff/675ddc4d704f4cde0bc72244263a9efbb0d32cb8
Modified Files
--------------
src/bin/pgbench/pgbench.h | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
From | Date | Subject | |
---|---|---|---|
Next Message | Andres Freund | 2025-09-12 14:19:32 | pgsql: ci: openbsd: Increase RAM disk's size |
Previous Message | Peter Eisentraut | 2025-09-12 06:05:45 | pgsql: Default to log_lock_waits=on |