search

OpenObservatory / ooni-pipeline.backup

[DEPRECATED] This is the ooni pipeline for doing data analytics on the collected reports
8 stars 6 forks source link

Ticket 13640 #11

Closed oripka closed 9 years ago

oripka commented 9 years ago

i took a look at 13640, here is a proposal to fix it: https://github.com/olliwolli/ooni-pipeline/commit/39afda26d48f092b08f4a8479053c0e65b048d8b . I didn't find any check to make sure that the control and experiment measurements actually are for the same bridge based on a hash. The code only checks if the time is is close. Maybe I overlooked something or this is intentional?

hellais commented 9 years ago

Ok so it appears that no checks on the input are done.

Here is a transcript from the #ooni channel:

13:59 < hellais> oh yes it looks like you are right
14:00 < hellais> kudrom, asn: can you confirm what otr is saying?
14:00 < hellais> that is that we don't check to see if the control measurement we are checking against is for the specific bridge measurement we are looking at?
15:24 < kudrom> hellais, otr: i think otr is right
15:24 < kudrom> currently we don't check the input of the measurements
15:26 < kudrom> i think it can be done in add_status_field with something like filtered_controls = filter(lambda x: x.get_test_name() == self.get_test_name(), controls)
15:26 < asn> i think otr is right too.
15:26 < kudrom> and passing the filtered_controls to find_closest
hellais commented 9 years ago

Ok I will proceed to merge this and rely on the other PR to fix the problem described above.