| 9 Oct 2021 |
das_j | In reply to @vcunat:matrix.org I think so, but in any case you can click on the build from which it was cached and restart that one. But wouldn't that possibly rebuild non-current builds? | 15:32:00 |
das_j | * But wouldn't that possibly build non-current builds? | 15:32:12 |
Vladimír Čunát | In reply to @janne.hess:helsinki-systems.de But wouldn't that possibly build non-current builds? It could, but typically we do have a separate job for non-trivial builds, except e.g. stdenv bootstrapping. | 15:32:48 |
hexa | In reply to @hexa:lossy.network we might need to revert the python39 bump on staging due to some rather interesting breakages in aiohttp et al waiting for feedback on https://github.com/NixOS/nixpkgs/pull/141105 from jon or fridh | 15:45:54 |
hexa | (This is basically what blocks the staging-small jobset) | 15:46:27 |
hexa | * (This is basically what unblocks the staging-small jobset) | 15:46:33 |
lukegb (he/him) | https://user-images.githubusercontent.com/11718525/135937807-fd3e0fd2-a31a-47a4-90c6-b0bb1d0704d4.png heh, cute | 16:45:12 |
lukegb (he/him) | ah, https://github.com/python/cpython/commit/c967bd523caabb05bf5988449487d7c1717f3ac6 | 16:49:49 |
hexa | yup, will be part of 3.9.8 afaict | 16:57:33 |
hexa | well, let me try to fetchpatch that one | 16:57:50 |
hexa | as soon as I've fixed mailman 😓 | 16:58:23 |
hexa | oh, which is now blocked by that python issue. lmao | 16:58:42 |
lukegb (he/him) | hahaha | 17:01:38 |
lukegb (he/him) | an infinite fractal of brokenness | 17:01:45 |
hexa | two hours in … soon™ | 18:54:34 |
hexa | nah, only a partial fix 😕 | 19:25:36 |
hexa | In reply to @vcunat:matrix.org Anyway, let's do a new staging-small eval: https://hydra.nixos.org/eval/1712002 restarted with python39 revert | 22:55:24 |
| 10 Oct 2021 |
hexa | I think python-unstable might become ready tonight for merge into staging | 16:53:48 |
jonringer | I'm going to do some fixes as well, just give me a few hours :) | 20:56:25 |
jonringer | actually, if it's going into staging, then just merge when ready | 20:58:24 |
hexa | jonringer: I think I'm about done. | 23:28:06 |
jonringer | I'll just do fixes on staging next | 23:30:04 |
hexa | wfm | 23:33:16 |
| 11 Oct 2021 |
hexa | and done. Well … bisecting ocrmypdf, but that's going to take a while. | 01:09:23 |
hexa | bisected to pybind11 2.8.0 bump | 14:09:24 |
| 13 Oct 2021 |
jonringer | hexa: The pikepdf tests which are failing (and then causing ocrmypdf to fail) because of timeouts. seems like pybind11 2.8.0 has some performance degradation. | 02:34:31 |
jonringer | tests/test_metadata.py:277:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
args = (1, 1, 1, 0, 0, 0), kwargs = {}, initial_draws = 0
start = 1835476.350838395, result = None, finish = 1835476.600072148
internal_draw_time = 0, runtime = datetime.timedelta(microseconds=249234)
current_deadline = timedelta(milliseconds=200)
@proxies(self.test)
def test(*args, **kwargs):
self.__test_runtime = None
initial_draws = len(data.draw_times)
start = time.perf_counter()
result = self.test(*args, **kwargs)
finish = time.perf_counter()
internal_draw_time = sum(data.draw_times[initial_draws:])
runtime = datetime.timedelta(
seconds=finish - start - internal_draw_time
)
self.__test_runtime = runtime
current_deadline = self.settings.deadline
if not is_final:
current_deadline = (current_deadline // 4) * 5
if runtime >= current_deadline:
> raise DeadlineExceeded(runtime, self.settings.deadline)
E hypothesis.errors.DeadlineExceeded: Test took 249.23ms, which exceeds the deadline of 200.00ms
/nix/store/729144mrgipxk30s2c049w28783r203v-python3.9-hypothesis-6.23.2/lib/python3.9/site-packages/hypothesis/core.py:588: DeadlineExceeded
---------------------------------- Hypothesis ----------------------------------
Falsifying explicit example: test_random_dates(
year=1, month=1, day=1, hour=0, mins=0, sec=0,
)
__________________________ test_random_valid_docinfo ___________________________
[gw6] linux -- Python 3.9.6 /nix/store/kjj6088w5bv56pqnkkcgpv9cw1y8082p-python3-3.9.6/bin/python3.9
> ???
E hypothesis.errors.FailedHealthCheck: Data generation is extremely slow: Only produced 1 valid examples in 11.61 seconds (0 invalid ones and 0 exceeded maximum size). Try decreasing size of the data you're generating (with e.g. max_size or max_leaves parameters).
E See https://hypothesis.readthedocs.io/en/latest/healthchecks.html for more information about this. If you want to disable just this health check, add HealthCheck.too_slow to the suppress_health_check settings for this test.
tests/test_metadata.py:593: FailedHealthCheck
| 02:34:45 |
jonringer | we could dis able the tests, since we aren't in the business of maintaining performance metrics | 02:35:10 |
jonringer | * we could disable the tests, since we aren't in the business of maintaining performance metrics | 02:35:15 |
hexa | I hadn't seen pikepdf fail, only ocrmypdf, and I applied a patch for the later | 03:30:23 |