Horton

Re: Bump MIN_WINNT to 0x0600 (Vista) as minimal runtime in 16~

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Thomas Munro <thomas.munro@gmail.com>
Cc: Justin Pryzby <pryzby@telsasoft.com>, Andrew Dunstan <andrew@dunslane.net>, Daniel Gustafsson <daniel@yesql.se>, Juan José Santamaría Flecha <juanjo.santamaria@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2022-09-08T08:55:40Z
Lists: pgsql-hackers

Attachments

On Tue, Aug 30, 2022 at 01:29:24PM +1200, Thomas Munro wrote:
> This reminds me of 24c3ce8f1, which replaced a dlopen()-ish thing with
> a direct function call.  Can you just call all these functions
> directly these days?

Hmm.  Some tests in the CI show that attempting to call directly
MiniDumpWriteDump() causes a linking failure at compilation.  Anyway,
in the same fashion, I can get some simplifications done right for
pg_ctl.c, auth.c and restricted_token.c.  And I am seeing all these
functions listed in the headers of MinGW, meaning that all these
should work out of the box in this case, no?

The others are library-dependent, and I not really confident about
ldap_start_tls_sA().  So, at the end, I am finishing with the
attached, what do you think?  This cuts some code, which is nice:
 3 files changed, 48 insertions(+), 159 deletions(-)
--
Michael

Commits

  1. doc: Fix incorrect version list for Windows SDK

  2. Replace loading of ldap_start_tls_sA() by direct function call

  3. Replace load of functions by direct calls for some WIN32

  4. Cleanup more code and comments related to Windows NT4 (XP days)

  5. Make Windows 10 the minimal runtime requirement for WIN32