14:29 <mattip> summarizing the latest py3.8 lib-python run gives {'Ran': 10045, 'failures': 199, 'errors': 24, 'skipped': 1642}

14:30 <mattip> so under 200 failures

14:33 <mattip> the worst are test_ast and test_dis, which we should just skip

14:34 <mattip> then test_utf8_mode (9), test_concurrent_futures(7) and test_multibytecodec(7)

14:38 <mattip> the 24 errors are scattered all over, test_hashlib(3), test_sqlite(2), test_subprocess(2), test_zlib(2), test_ctypes(2), test_dict(2),

14:38 <mattip> test_dict might be worth looking at, it seems dict is slightly off

14:39 <mattip> but I think the bottom line is we can release this as-is

15:58 <mattip> about "make clean" as part of the buildbot run: how would the buildbot know the /tmp directory to run "make clean" in?

15:58 <mattip> maybe we could add a "cleanup" command line argument to translation that would do this (and also delete any other temporary directories created)

16:14 <mattip> at the same time I would like to remove the hack that during translation we run build_cffi, it originally started before we added that to packaging.py

16:14 <mattip> and is now redundant

20:21 <ceridwen> I'm trying to run vmprof in lines mode using --filter on a function, but seeing nothing in the hits/% hits columns though this is the function according to the tree and flat views where the program is spending most of the time. I'm assuming I've hit some limitation of vmprof or lines mode, is there an easy way to find out what that is?