Re: pgsql: Redesign handling of SIGTERM/control-C in parallel pg_dump/pg_re

Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com> writes:
> On 6/2/16 1:28 PM, Tom Lane wrote:
>> Redesign handling of SIGTERM/control-C in parallel pg_dump/pg_restore.

> These changes introduced several new compiler warnings under fortify rules:

> parallel.c: In function sigTermHandler:
> parallel.c:556:9: warning: ignoring return value of write, declared
> with attribute warn_unused_result [-Wunused-result]

Hm, interesting --- I copied that write_stderr() macro from psql/common.c
and figured it was good. But now that I look, only the uses of it in the
Windows code path there are straightforward; on the Unix side we have

rc = write_stderr("Cancel request sent\n");
(void) rc; /* ignore errors, nothing we can do here */

which evidently was done to shut up exactly this type of overly-nannyish
warning. I'm thinking we need to hide that dead-chicken-waving in the
macro itself. Will deal with it tomorrow.

regards, tom lane