<rjo>
whitequark: ok. how can we fix the buildbot? in the long term everybody just fiddling and bolting onto it won't really work.
<whitequark>
rjo: fix as in have a single source of authority?
<rjo>
whitequark: currently fix it so it builds bitstreams again. but in the long term we need to at least track what we do and why we did it and what issues we solved how.
<whitequark>
I have no idea how to fix this failure, the last time sb0 twiddled it somehow I think
<whitequark>
in the long term, sure, I can set up ansible
<whitequark>
it's a tire fire alright, but it's a significantly smaller tire fire than chef or puppet, and only moderately revolting
<sb0>
this problem existed before I touched it, it's been happening sporadically for many months
<sb0>
and yes, this kind of buildbot non-determinism is obnoxious
sb0 has quit [Quit: Leaving]
<whitequark>
it's just the usual conda garbage, no?
<whitequark>
if you want me to rollback the VM before every build, no problem, I can do that almost trivially
jbqubit_ has joined #m-labs
sb0 has joined #m-labs
<sb0>
VM rollback does sound like a good idea
<sb0>
or just reset the conda stuff
<sb0>
I think there are parts that are not built in a VM?
sb0 has quit [Quit: Leaving]
sb0 has joined #m-labs
sb0 has quit [Quit: Leaving]
sb0 has joined #m-labs
rohitksingh has joined #m-labs
sb0 has quit [Quit: Leaving]
sb0 has joined #m-labs
rohitksingh has quit [Quit: Leaving.]
<whitequark>
I don't think so?
<whitequark>
I'm not sure but I don't think so
<whitequark>
oh, hang on
<whitequark>
ARTIQ is not built in a VM, we use VMs only for the rest of the conda packages
<whitequark>
ameliorating this requires an incredibly annoying dance with USB forwarding every time we're flashing something
<whitequark>
I'm not ready to debug that
<whitequark>
(we need the dance because kc705 is used both from the host VM and the one-off build VM...)
<whitequark>
(I mean host OS)
<sb0>
anyway right now the priority should be the TCP bug and other 3.0 issues
<whitequark>
ok...
cr1901_modern has joined #m-labs
<GitHub177>
[smoltcp] batonius opened pull request #15: Factor out the "raw_socket" and "tap_interface" features (master...split_std) https://git.io/vHF2d