[ooni-dev] http_requests test single responses

[jony-port at bitmessage.ch]

On Mon, 11 Aug 2014 20:06:52 +0200
Arturo Filastò <art at torproject.org> wrote:

> On 8/11/14, 6:34 PM, jony-port at bitmessage.ch wrote:
> > Hi,
> > 
> > On Thu, 31 Jul 2014 14:05:49 +0200
> > Arturo Filastò <art at torproject.org> wrote:
> > 
> >> On 7/24/14, 2:52 PM, jony-port at bitmessage.ch wrote:
> >>>
> >>> It seems that the single response of the http_requests requests
> >>> test results in 'body_length_match' and 'headers_match' fields set
> >>> to true.
> >>>
> >>
> >> Yes that is correct. Those values are initialized to none and if a
> >> response is received for both control and experiment they are set
> >> to either true or false.
> > 
> > But when there is only one request this should never be as true,
> > since there is only one request, thus nothing to compare with or I
> > miss anything?
> > 
> 
> Wait, the http_requests test should be doing 2 requests. One is over
> your network connection, the other is over tor. Looking back at the
> report you sent it seems like you are only sending 1 request.
> 
> Did you make any changes to the http_requests test and which version
> of ooni-probe are you running?

I have seen this issue in reports made from various probes most of
them using the official debian package or the latest git release.

No changes were made to the http_requests test.

> Can you enable debug output and paste somewhere the output of it (you
> need to edit ~/.ooni/ooniprobe.conf to set advanced->debug: true)

I cannot reproduce this on my system. These reports were gathered at
some point from different systems, most of them are on the main reports
archive (https://ooni.torproject.org/reports/0.1/).

In any case, I guess it would be better to check that the headers exist
before calling getDiff
(https://gitweb.torproject.org/ooni-probe.git/blob/HEAD:/ooni/nettests/blocking/http_requests.py#l94)