Simone Basso

Uhm, I think this needs more investigations and perhaps more data points. Next time a build is stuck on travis, I'll gather more data and report it here. Meanwhile, let's...

[Another build stalled](https://travis-ci.org/measurement-kit/measurement-kit/jobs/512244535). The diff between this build and the successful one presented above is: ```diff --- STUCK 2019-03-28 00:17:41.000000000 +0100 +++ ORIG_GOOD 2019-03-28 00:16:34.000000000 +0100 @@ -1,3 +1,4 @@...

[Another build stalled](https://travis-ci.org/measurement-kit/measurement-kit/jobs/574373693). The diff between this build and the successful one presented above is: ```diff $ diff -u ORIG_GOOD STUCK --- ORIG_GOOD 2019-08-20 16:46:18.000000000 +0200 +++ STUCK 2019-08-20 16:46:12.000000000...

I've looked into the measurement. Here's an excerpt: ```json { "queries": [ { "engine": "system", "resolver_hostname": null, "query_type": "A", "hostname": "edge-mqtt.facebook.com", "answers": [], "failure": "dns_resolver_error", "resolver_port": null }, { "engine":...

I discussed this further with @hellais. The main issue is that there are some fields that are supposed to be toplevel and that shall always be present, regardless of what...

The best way to handle this is probably using Jafar.

This is a similar case, but for WhatsApp: https://explorer.ooni.org/measurement/20200219T021904Z_AS9916_BHgtaurQnDpvgO9LWgNJMCrDyyfLuwMMsAKR28GPZYBOm3QTQY Thanks @FedericoCeratto!

There is quite a scary logic for deciding when something is blocked or accessible. I believe I should refactor quite heavily this test implementation to simplify it. I'll do this...

Moved to https://github.com/ooni/probe-engine/issues/579#issuecomment-660662613

Actually, I'd like to keep this open, since it's a bug of the MK implementation.