Clean up minor inconsistencies in pg_attribute_printf() usage.

For some reason we'd never decorated pg_v*printf() with
pg_attribute_printf() annotations.  There is a convention for
how to label va_list-using printf functions (write zero for the
second argument), and we use that liberally elsewhere in the
code, but these core functions lacked it.  It's not clear how
much useful checking the compiler can do for calls of these,
but we might as well add the annotations.

Also, sync win32security.c's log_error() with our normal convention
that pg_attribute_printf must be attached to a function's declaration
not definition.  Apparently this file is only compiled with compilers
that aren't picky about that, but still it'd be better to be
consistent.

No back-patch since there's little reason to think we would catch
anything.

Discussion: https://postgr.es/m/3492412.1663283395@sss.pgh.pa.us

0 files changed, 0 insertions, 0 deletions