On Sun, Jan 3, 2021 at 9:49 AM Fabien COELHO <coelho@cri.ensmp.fr> wrote:
> > Just for fun, the attached 0002 patch is a quick prototype of a
> > replacement for that stuff that seems to work OK on a Mac here. (I'm
> > not sure if the Windows part makes sense or works.)
>
> Thanks! That will definitely help because I do not have a Mac. I'll do
> some cleanup.
I think the main things to clean up are:
1. pthread_barrier_init() should check for errors from
pthread_cond_init() and pthread_mutex_init(), and return -1.
2. pthread_barrier_destroy() should call pthread_cond_destroy() and
pthread_mutex_destroy().
3 . Decide if it's sane for the Windows-based emulation to be in here
too, or if it should stay in pgbench.c. Or alternatively, if we're
emulating pthread stuff on Windows, why not also put the other pthread
emulation stuff from pgbench.c into a "ports" file; that seems
premature and overkill for your project. I dunno.
4. cfbot shows that it's not building on Windows because
HAVE_PTHREAD_BARRIER_WAIT is missing from Solution.pm.
As far as I know, it's OK and common practice to ignore the return
code from eg pthread_mutex_lock() and pthread_cond_wait(), with
rationale along the lines that there'd have to be a programming error
for them to fail in simple cases.
Unfortunately, cfbot can only tell us that it's building OK on a Mac,
but doesn't actually run the pgbench code to reach this stuff. It's
not running check-world on that platform yet for the following asinine
reason:
connection to database failed: Unix-domain socket path
"/private/var/folders/3y/l0z1x3693dl_8n0qybp4dqwh0000gn/T/cirrus-ci-build/src/bin/pg_upgrade/.s.PGSQL.58080"
is too long (maximum 103 bytes)