On Wed, Nov 12, 2025 at 12:10:03AM +0100, Bruno Haible wrote:

> I also observe similar failures in 'test-cond' and 'test-cnd'.
> 
> Changing the type of 'cond_timed_out' from 'int volatile' to
> 'struct atomic_int' does not help. But this patch fixes the test failures:

Thanks a lot. I'd like to understand this better.

Am I right to think that this is not really an architecture-specific
problem but more like a consequence of the machine being "slow"?

Also, how do we determine that using 3 seconds instead of 2 is a good fix?
Is that just the smallest increment that made your 9 in 100 failure
rate to become unmeasurable? (I guess 0 after trying a lot of times)

Note: The failure is also random here, as I clicked on the retry
button and then it worked:

https://buildd.debian.org/status/fetch.php?pkg=gettext&arch=sparc64&ver=0.23.2-1%7Eexp1&stamp=1762914447&raw=0

(But I'll apply the patch anyway for unstable)

Thanks.

Reply via email to