Postgres Pro
Downloads
Home   >   mailing lists

Re: Can't compile with profiling after BRIN autosummarization - Mailing list pgsql-hackers

From Jeff Janes
Subject Re: Can't compile with profiling after BRIN autosummarization
Date
Msg-id CAMkU=1wrymb0VveZrcQz3V=AXNRuPiYMVYQO_aMdPgm0yR+2Zg@mail.gmail.com
Whole thread Raw
In response to Can't compile with profiling after BRIN autosummarization  (Jeff Janes <jeff.janes@gmail.com>)
List pgsql-hackers
Tree view
On Mon, Apr 3, 2017 at 9:31 AM, Jeff Janes <jeff.janes@gmail.com> wrote:
make maintainer-clean > /dev/null ; ./configure  --enable-profiling > /dev/null &&  make -j8 >/dev/null

In file included from ipc.c:29:
../../../../src/include/postmaster/autovacuum.h:73: error: expected declaration specifiers or '...' before 'BlockNumber'
make[4]: *** [ipc.o] Error 1
make[3]: *** [ipc-recursive] Error 2
make[3]: *** Waiting for unfinished jobs....
make[2]: *** [storage-recursive] Error 2
make[2]: *** Waiting for unfinished jobs....
make[1]: *** [all-backend-recurse] Error 2
make: *** [all-src-recurse] Error 2


This git bisects down to:

commit 7526e10224f0792201e99631567bbe44492bbde4
Author: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date:   Sat Apr 1 14:00:53 2017 -0300

    BRIN auto-summarization

The lines are:

extern void AutoVacuumRequestWork(AutoVacuumWorkItemType type,
                      Oid relationId, BlockNumber blkno);

I have no idea why --enable-profiling would change the way this line is seen.

gcc version 4.4.7 20120313 (Red Hat 4.4.7-17) (GCC)

A newer gcc gave a better error message which lead me to add the line:

#include "storage/block.h"         /* for typedef BlockNumber */
 
into autovacuum.h, which fixes the problem.

I don't know if that is the correct fix, or why it is only needed under profiling.

Cheers,

Jeff

pgsql-hackers by date:

Previous
From: Kevin Grittner
Date:
Subject: Re: GSoC 2017 Proposal for "Explicitly support predicatelocks in index access methods besides btree"
Next
From: Keith Fiske
Date:
Subject: Re: pg_partman 3.0.0 - real-world usage of nativepartitioning and a case for native default