Update of task #7050 (group hurd):
Summary: implement shared semaphores and mutexes, implement
sem_open/semget => implement shared barriers, rwlocks, conds
Wiki-like text discussion box: As a starter, maybe sem_open() would be easier,
because it seems like it could map very well to a /servers/sem/ directory
where named semaphores would live.
=> sem_open and mutex_init were done, remaining work needed for barriers,
rwlocks, conds.
_______________________________________________________
Follow-up Comment #3:
As my comment said, shared semaphore in terms of sem_open did get implemented,
with shared memory support, so postgresql can indeed use them. The claim
"PostgreSQL can't run on GNU/Hurd apparently, for lack of shared sempahores
(sic)" is obviously misinformed, and voluntarily breaking the build won't help
for sure.
Help welcome to check what actually is a problem with postgresql on GNU/Hurd,
at least it's really not a lack for shared semaphores, they really are there.
As for a data point, see
https://buildd.debian.org/status/fetch.php?pkg=postgresql-16&arch=hurd-i386&ver=16.1-1%2Bb1&stamp=1705091712&raw=0
# 2 of 215 tests failed.
It doesn't look that broken to me...
_______________________________________________________
Reply to this item at:
<https://savannah.gnu.org/task/?7050>
_______________________________________________
Message posté via Savannah
https://savannah.gnu.org/
