05:55 <fijal> mattip: hey

07:07 <mattip> fijal: pong

07:11 <mattip> there is a new failure on py3.7 in lib-python.3.test.test_poplib, seems to be due to something in _ssl _read_no_buf

07:56 <mattip> untranslated hpy tests are timing out on windows, I think it is because the make_module() call is very slow

07:57 <mattip> some of the tests call make_module() in a for loop rather than combining a number of functions into a single module and compiling once

08:11 <mattip> overall, running pypy2.7 pytest.py pypy\module\_hpy_universal\test takes over an hour

08:34 <mattip> running extra_tests also takes an hour, where on default it takes 24 minutes

10:22 <mattip> we run the cffi tests and hpy tests both in -A and in extra_tests. On linux it adds ~20 minutes, and on windows over 90 minutes

10:23 <mattip> why do we need both?

10:37 <mattip> fijal: the new _ssl code is causing lib-python/3/test_ssl to hang and other tests to fail

11:43 <mattip> I merged hpy-0.0.3 to py3.7

11:44 <mattip> now trying to figure out openssl-3.0.0 before the next release candidate

14:29 <ctismer> Hi mattip, do you have plans about the `PyType_Modified` issue?

14:30 <ctismer> mattip: I'm just asking because I have an ugly patch ready, but if I can avoid it...

14:34 <mattip> I was waiting for some kind of response from CPython, otherwise it will remain difficult to discover.

15:39 <ctismer> mattip: I was talking about `PyType_Ready`s type finishing and your thoughts to invalidate this cache when noticed by `PyType_Modified`. What role does CPython's response play there?

16:16 <ronan> mattip: the hpy tests don't run under -A