< gkrizek>
I requested an invite to the Bitcoin Core Slack and haven't heard anything. How can I get an invite? Is IRC or Slack used more?
< booyah>
gkrizek: the regular meetings, are happening here on irc
< gkrizek>
booyah: Thanks! I'll probably just stick to IRC then
< luke-jr>
gkrizek: I don't think any Core devs actually use the Slack at this point
< gkrizek>
luke-jr: Ha, thanks for the heads up.
< sipa>
no, it's just a community effort
< gkrizek>
Looking at the master branch in bitcoin/bitcoin, I see a lot of PRs are merged with failing tests. Why is this? Are the tests just flaky?
< sipa>
the appveyor tests are experimental, and we ignore them for merging
< sipa>
unfortunately we can't prevent it from marking the github PR as invalid
< sipa>
*failed
< gkrizek>
Understood. But it seems like Travis fails at a similar frequency...
< gkrizek>
sipa: I'm new and want to start contributing. I'm just trying to understand the flow and requirements. Mainly curious how important/accurate those test suites are.
< gwillen>
travis is generally pretty reliable
< gwillen>
I think travis fails are typically treated as real, although I admit I saw one just today that I'm pretty sure was spurious
< sipa>
there are occasionally failures due to timeouts or so
< gwillen>
but in that case all but one platform showed as passing, which I think is a good clue
< gwillen>
you can also inspect the output
< gwillen>
so I think the answer is "the tests must actually pass, but spurious CI failures that aren't actually the result of failing tests happen sometimes"
< gkrizek>
Ok, awesome. Thanks for the explanation!
< phantomcircuit>
gwillen, travis randomly fails too
< phantomcircuit>
it seems like the tests are much more expensive to run than they used to be (which is fine, but breaking)