Re: WIP: Rework access method interface - Mailing list pgsql-hackers

On Mon, Dec 14, 2015 at 5:26 PM, Petr Jelinek <petr@2ndquadrant.com> wrote:

I went over this.

I get these compiler warning about unused variables in the validation functions:
brin.c: In function ‘brinvalidate’:
brin.c:94:6: warning: variable ‘keytype’ set but not used [-Wunused-but-set-variable]
      keytype;
      ^
ginutil.c: In function ‘ginvalidate’:
ginutil.c:86:6: warning: variable ‘keytype’ set but not used [-Wunused-but-set-variable]
      keytype;
      ^
gist.c: In function ‘gistvalidate’:
gist.c:101:6: warning: variable ‘keytype’ set but not used [-Wunused-but-set-variable]
      keytype;
      ^
hash.c: In function ‘hashvalidate’:
hash.c:103:6: warning: variable ‘keytype’ set but not used [-Wunused-but-set-variable]
      keytype;
      ^
nbtree.c: In function ‘btvalidate’:
nbtree.c:134:6: warning: variable ‘keytype’ set but not used [-Wunused-but-set-variable]
      keytype;
      ^
nbtree.c:133:6: warning: variable ‘intype’ set but not used [-Wunused-but-set-variable]
      intype,
      ^
spgutils.c: In function ‘spgvalidate’:
spgutils.c:88:6: warning: variable ‘keytype’ set but not used [-Wunused-but-set-variable]
      keytype;
      ^

These look like copy-pastos of boilerplate.

Another note is that amvalidate SQL interface is not documented anywhere. I know it's mainly meant for regression tests and we for example don't document hashing functions but it's something to think about/discuss maybe.

What do you think about System Administration Functions

Other than that I am happy with the patch.