Since in OONI we typically use a three class system (i.e. accessible,
anomaly, or blocking), I ...

Avoids requesting several times in a row for the control SNI, which
is frankly a waste of time ...

Part of https://github.com/ooni/probe-engine/issues/388

Makes the documentation more beautiful.

Part of https://github.com/ooni/probe-engine/issues/302.

Part of https://github.com/ooni/probe-engine/issues/302

This is marked API CHANGE because now...

As mentioned in https://github.com/ooni/probe-engine/issues/302, this split of the experiment an...

This is part of https://github.com/ooni/probe-engine/issues/302.

I feel comfortable with remo...

Closes https://github.com/ooni/probe-engine/issues/338, because I have
added code and unit test...

Part of https://github.com/ooni/probe-engine/issues/302

The vendoring has omitted files we don't care about. The source code is
not changed. The README...

This is part of https://github.com/ooni/probe-engine/issues/302

Closes https://github.com/ooni/probe-engine/issues/348

When registering, we don't need to lookup the country and the ASN. Desktop
clients only use orc...

Also part of https://github.com/ooni/probe-engine/issues/334

* internal/resources: update assets to 20200225143707

* mmdblookup.go: adapt to db-ip.com's d...

It seems we cannot access this service without using domain
fronting and we need to use the a24...

We were too strict and this means coding the Android app to
support MK and probe-engine at the ...

The matter is better explained in the newly added code comment. I feel
okay removing the code, ...

See https://github.com/ooni/probe-engine/pull/354

See https://github.com/ooni/probe-engine/pull/347

Developed as part of https://github.com/ooni/probe-engine/issues/339.

This is useful to have ...

While there, make `resources` part of `internal` since it shouldn't be exposed.

While there, ...

I want to build an `.aar` file that only includes experiments written in Go such that I can link...

See https://github.com/ooni/probe-engine/issues/316

experiment: say data_format_version 0.2.0

This avoids confusing the pipeline. We use an annotation
to report the real data_format_version.

dependencies: update bolt and goselect because of other arch support

Signed-off-by: Jan Pavlinec <[email protected]>

<facepalm>

Related to #314

Closes https://github.com/ooni/probe-engine/issues/304

Implements https://github.com/ooni/spe...

Start replacing ./httpx with github.com/ooni/netx

I did determine that it was enough to rely on the retry behaviour
of the default resolver, back ...

We really don't need it there. We can use http.DefaultClient directly.

We now have full coverage of this package.

No big deal, except that we are not distracted by coverage not being
100% here so we can focus o...

As part of the work to fully migrate to netx, let's also dismantle
the httpx directory which we ...

Closes #265

See https://github.com/ooni/sysadmin/commit/bf082ff0953d72bdf67773b2527716e50646908e

Part of ...

Closes https://github.com/ooni/probe-engine/issues/292

No real change except that I like new instructions a bit more.

probe-engine 0.5.0 contains updated dependencies

I've improved my understanding of how to do that and have
changed the related documentation acco...

The command I've used is the one listed in the README.md:

```
for name in `grep -v indirect go....

We're removing scoreboard since support for it has been previously
removed from netx in https://...

Closes https://github.com/ooni/probe-engine/issues/226 and #276.

* psiphon: add TLS handshake data

This is consistent w/ psiphon's spec v0.3.2

* telegram: ...

Part of #12.

Part of #12. Code from #26.

We use port numbers in netx because this is a way to make sure
that we can bind together HTTP a...

This is also functional to #12. This follows up from #218.

While there, also ignore the `.vs...

No functional change because the default is still that we use
the current time, if beginning is...

Related to https://github.com/ooni/spec/pull/174, which was
in turn motivated by https://github...

I'd like to include the events to the output of #90.

Turns out that my understanding of data races was imprecise.

While there, apply some cosmetic...

* oonitemplates: implement OBFS4Connect template

There will later be opportunity to improve t...

* oonitemplates.go: allow skipping TLS verification

We're going to need this in #2 to connect...

When there's an error, HTTPRoundTripDone is emitted but we should
not attempt printing headers ...

Before this diff, we could not unmarshal when the body was binary
as I discovered while working...

This way we pass the same argument type to all functions, which
is convenient when we have HTTP...

Closes #191. Necessary to export predictable mobile API.

This was not implemented up until now, but actually is better
to have it, because #2 can also i...

So, it's close to oonitemplates. This is yak shaving that I am
doing in preparation for #90.

This template is going to be useful to implement #2.

We are going specifically to use it to v...

These are two experimental packages that are too bound to this repo
to be kept into the other r...

Closes #114

Closes #178

This patch changes the way in which we implement the kvstore used
to keep track of orchestra, t...

Using filepath.Join leads to wrong results on Windows where the
`\` path separator is used inst...

orchestra: store authentication information

While in Go strings can be binary, this is not the case with java
and sqlite3. So, be very clear...

Just an observation based on reading the code of the statefile
in memory currently being used by...

By default the key-value store is in memory. We will be able to
easily override it for ooni/prob...

The abstraction is already captured by the generic key-value store.

Part of #164. We are now able to use a generic key-value store
for the orchestra statefile imple...

As mentioned in https://github.com/ooni/probe-engine/issues/112#issuecomment-566958248,
I have ...

* psiphon: simplify now that we get config from orchestra

Closes #168

* Remove error we do...

The previous fix in #174 was correct, but not enough. The correct fix is
to also include the Te...

Closes #167

Closes #109

To this end, update to neubot/[email protected]

Closes #155.

Closes #169

This should close #142. Because maps have no stable sorting in Go, we
sometimes pick the succes...

Suggested by @mirokuratczyk when we discussed the experiment
in a recent chat on Slack. Closes ...

Closes #146