Re: [RFC] building postgres with meson -v8

Hi,

On 2022-05-18 10:30:12 +0200, Peter Eisentraut wrote:
> Here are some more patches that clean up various minor issues.

I rebased the meson tree, squashed a lot of the existing commits, merged your
changes, and fixed a few more differences between autoconf and meson.

For me the difference in defines now boils down to:

- CONFIGURE_ARGS - empty in meson, not clear what to fill it with
- GETTIMEOFDAY_1ARG - test doesn't exist - I suspect it might not be necessary
- PACKAGE_STRING, PACKAGE_TARNAME - unclear if they should be implemented?
- AC_APPLE_UNIVERSAL_BUILD logic - which I don't think we need?
- pg_restrict is defined in a simplistic way
- "missing" a bunch of defines that don't appear to be referenced:
HAVE_FSEEKO
HAVE_GSSAPI_GSSAPI_H
HAVE_INTTYPES_H
HAVE_LDAP_H
HAVE_LIBCRYPTO
HAVE_LIBLDAP
HAVE_LIBM
HAVE_LIBPAM
HAVE_LIBSSL
HAVE_LIBXML2
HAVE_LIBXSLT
HAVE_MEMORY_H
HAVE_PTHREAD
HAVE_PTHREAD_PRIO_INHERIT
HAVE_STDINT_H
HAVE_STDLIB_H
HAVE_STRING_H
HAVE_SYS_STAT_H
HAVE_SYS_TYPES_H
HAVE_UNISTD_H
SIZEOF_BOOL
SIZEOF_OFF_T
STDC_HEADERS
- meson additional defines, seems harmless:
HAVE_GETTIMEOFDAY - only defined on windows rn
HAVE_SHM_UNLINK
HAVE_SSL_NEW
HAVE_STRTOQ
HAVE_STRTOUQ
HAVE_CRYPTO_NEW_EX_DATA
- a bunch of additional #undef's

Greetings,

Andres Freund