emma issueshttps://gitlab.torproject.org/tpo/anti-censorship/emma/-/issues2022-03-01T17:53:55Zhttps://gitlab.torproject.org/tpo/anti-censorship/emma/-/issues/7Capture blocking by "TCP works but SSL gets mangled"2022-03-01T17:53:55ZCecylia BocovichCapture blocking by "TCP works but SSL gets mangled"We had a user run emma in Belarus, where they had no access to default bridges, and received results that showed no directory authorities, relays, or bridges were being blocked. This could be because emma only performs a TCP connection a...We had a user run emma in Belarus, where they had no access to default bridges, and received results that showed no directory authorities, relays, or bridges were being blocked. This could be because emma only performs a TCP connection and the blocking is happening at the TLS layer.
See https://gitlab.torproject.org/tpo/anti-censorship/censorship-analysis/-/issues/40002
Can we expand these tests to capture this behaviour while still keeping the lightweight design?https://gitlab.torproject.org/tpo/anti-censorship/emma/-/issues/6Maybe try more than once if a connection timed out?2021-02-07T21:57:09ZPhilipp Winterphw@torproject.orgMaybe try more than once if a connection timed out?I tried to make sense of recent emma runs in Uganda (see tpo/community/outreach#40007) and noticed that it's difficult to tell if a connection failed because of reliability issues or because of censorship. Emma currently tries once to [e...I tried to make sense of recent emma runs in Uganda (see tpo/community/outreach#40007) and noticed that it's difficult to tell if a connection failed because of reliability issues or because of censorship. Emma currently tries once to [establish a TCP connection](https://gitlab.torproject.org/tpo/anti-censorship/emma/-/blob/2ed24be87ec7b6988ebc3c3723e53a2f7f17f89d/tests.go#L36) to a target and if that fails, it labels the target as unreachable. It's up to the operating system to determine the number of TCP retransmissions. Let's teach emma to try more than once to establish a TCP connection to a target. This will increase the test time a little bit but I think the improved clarity is worth the additional wait time.https://gitlab.torproject.org/tpo/anti-censorship/emma/-/issues/5Incorporate emma into OONI2024-02-29T15:21:09ZPhilipp Winterphw@torproject.orgIncorporate emma into OONIMaria once suggested adding emma to OONI. Both are written in Go, so it may not be a terribly complex endeavour. I'll have a chat with Simone to get a better sense of what this would entail.
The big benefit of having emma in OONI is th...Maria once suggested adding emma to OONI. Both are written in Go, so it may not be a terribly complex endeavour. I'll have a chat with Simone to get a better sense of what this would entail.
The big benefit of having emma in OONI is that we would get significantly more measurements and we no longer need to expect users to be able to compile programs.https://gitlab.torproject.org/tpo/anti-censorship/emma/-/issues/4Build Encore-style Web-based censorship measurement tests2022-02-04T18:43:43ZPhilipp Winterphw@torproject.orgBuild Encore-style Web-based censorship measurement testsEmma's source code must be compiled before one can run it, which makes it difficult for non-technical users to run emma. (Granted, we can cross-compile a .exe for Windows users but the resulting binary isn't signed and we shouldn't encou...Emma's source code must be compiled before one can run it, which makes it difficult for non-technical users to run emma. (Granted, we can cross-compile a .exe for Windows users but the resulting binary isn't signed and we shouldn't encourage people to run random executables.)
I propose to experiment with a Web-based version of emma. The idea is to point a user to example.com/emma/, which performs the same tests as our compiled executables but it uses JavaScript to get the job done. Here are a few projects that can serve as inspiration:
* Encore: https://censorbib.nymity.ch/#Burnett2015a
* CensorSweeper: https://lists.torproject.org/pipermail/tor-dev/2012-July/003851.html
* I believe that NetBlocks is doing something similar: https://netblocks.orghttps://gitlab.torproject.org/tpo/anti-censorship/emma/-/issues/3Increase test timeout2020-09-01T01:16:02ZPhilipp Winterphw@torproject.orgIncrease test timeoutEmma currently labels a test as failed if it didn't get a response within three seconds. That's a rather low timeout. We should probably change it to five seconds, if not more.Emma currently labels a test as failed if it didn't get a response within three seconds. That's a rather low timeout. We should probably change it to five seconds, if not more.https://gitlab.torproject.org/tpo/anti-censorship/emma/-/issues/2accounts.google.com is localised, which fails test for non-English environments2020-06-26T21:52:38ZPhilipp Winterphw@torproject.orgaccounts.google.com is localised, which fails test for non-English environmentsWe currently test if accounts.google.com is reachable by looking for the string "Sign in". In non-English environments, there is no such string. We need to be smarter about how we conduct this test.We currently test if accounts.google.com is reachable by looking for the string "Sign in". In non-English environments, there is no such string. We need to be smarter about how we conduct this test.https://gitlab.torproject.org/tpo/anti-censorship/emma/-/issues/1Facilitate cross-compilation for OS X2020-06-22T17:28:17ZPhilipp Winterphw@torproject.orgFacilitate cross-compilation for OS XWe currently only compile emma for Windows and Linux. People on OS X want to run it too.We currently only compile emma for Windows and Linux. People on OS X want to run it too.Philipp Winterphw@torproject.orgPhilipp Winterphw@torproject.org