Re: palloc unification - Mailing list pgsql-hackers

From Alvaro Herrera
Subject Re: palloc unification
Date
Msg-id 20130211215814.GD4737@alvh.no-ip.org
Whole thread Raw
In response to Re: palloc unification  (Alvaro Herrera <alvherre@2ndquadrant.com>)
Responses Re: palloc unification  (Alvaro Herrera <alvherre@2ndquadrant.com>)
List pgsql-hackers
Here's a more finalized version of this.  There are two main interesting
changes here:

1. postgres_fe.h includes common/fe_memutils.h
   This means all frontend programs get the #include without having to
   do it explicitely by themselves.  postgres.h includes utils/palloc.h
   which I used as precendent for this.

2. fe_memutils.h includes utils/palloc.h
   This means all frontend programs get the palloc() et al prototypes.

Some other things of note:

a. pg_upgrade was using postgres.h all over the place instead of
   postgres_fe.h.  There's no reason for this AFAICS so I changed it.
   No warnings, make check passes.
b. pg_resetxlog is the only frontend place that uses pg_malloc() and
   which needs to #include postgres.h instead of postgres_fe.h, so it's
   the only place that needs to include fe_memutils.h directly.
   (pg_controldata is the other place that includes postgres.h, but it
   doesn't need the pg_malloc etc declarations).
c. I added the MSVC bits.  I tested that most of it works, but the
   various regress executables as well as zic failed to build due to
   lack of libpgcommon at link time.  I think I fixed it; I'm waiting on
   new tests to run.  (This patch is the fixed version).

--
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

Attachment

pgsql-hackers by date:

Previous
From: Phil Sorber
Date:
Subject: Re: [PATCH] Add PQconninfoParseParams and PQconninfodefaultsMerge to libpq
Next
From: Andrew Dunstan
Date:
Subject: Re: JSON NULLs