Re: s_lock_test no longer works
Hi, On 2024-01-24 15:05:12 -0500, Tom Lane wrote: > Andres Freund writes: > > On 2024-01-24 12:14:17 +0100, Alvaro Herrera wrote: > >> I do wonder if we want to keep this around, given that it's been more > >> than one year broken and nobody seems to have noticed, and the Meson > >> build does not support the test as a target. > > > Perhaps we should just make the test built and run by default instead? > > OTOH, > > regress.c:test_spinlock() actually covers about as much as the standalone > > test... > > If your spinlocks aren't working, it's unlikely you'll get as far as > being able to run test_spinlock(). I think the standalone test does > have some value; it's just that it's not needed very often these days. As long as the uncontended case works, you can get surprisingly far... But still, fair enough. If so, I think we should just rig things so the standalone test gets built and run by default. It's not like that'd be a measurably expensive thing to do. Greetings, Andres Freund
Re: s_lock_test no longer works
Andres Freund writes: > On 2024-01-24 12:14:17 +0100, Alvaro Herrera wrote: >> I do wonder if we want to keep this around, given that it's been more >> than one year broken and nobody seems to have noticed, and the Meson >> build does not support the test as a target. > Perhaps we should just make the test built and run by default instead? OTOH, > regress.c:test_spinlock() actually covers about as much as the standalone > test... If your spinlocks aren't working, it's unlikely you'll get as far as being able to run test_spinlock(). I think the standalone test does have some value; it's just that it's not needed very often these days. regards, tom lane
Re: s_lock_test no longer works
Hi, On 2024-01-24 12:14:17 +0100, Alvaro Herrera wrote: > I do wonder if we want to keep this around, given that it's been more > than one year broken and nobody seems to have noticed, and the Meson > build does not support the test as a target. Perhaps we should just make the test built and run by default instead? OTOH, regress.c:test_spinlock() actually covers about as much as the standalone test... Greetings, Andres Freund
Re: s_lock_test no longer works
Alvaro Herrera writes: > I do wonder if we want to keep this around, given that it's been more > than one year broken and nobody seems to have noticed, and the Meson > build does not support the test as a target. The last time it was broken, it took us multiple years to notice, too. I'm not sure that that's a reason to remove the test scaffolding, though. You'd probably only really use it to smoke-test some new spinlock assembly code, and how often does anyone do that anymore? regards, tom lane
