| Commit message (Collapse) | Author | Age |
| | |
|
| |\ |
|
| | | |
|
| | | |
|
| | |
| |
| |
| |
| | |
Additionally, give them nicer unique names, and warn if we
accidentally double-add a test or something.
|
| | |
| |
| |
| | |
This should prevent multiple in-process runs from interfering
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | |
| |
| |
| |
| | |
The default client/responder pair is no longer source/sink, but
echoclient/echoserver.
|
| | |
| |
| |
| |
| | |
We'd like to be able to change the client and responder behavior,
but we need to keep both parts synchronized as we do so.
|
| | | |
|
| | |
| |
| |
| |
| |
| | |
Now data is generated by a DataSource type, and consumed by a
DataChecker type that compares its incoming data against the data
in a DataSource.
|
| | |
| |
| |
| |
| |
| |
| |
| | |
Previously we counted the number of successes and failures, and
compared them to the number of expected tests. That's fragile, if
we forget to count a test or if we count a success more than once.
Now we give every test a name, and we only allow tests to succeed or
fail once.
|
| | |
| |
| |
| |
| |
| |
| |
| |
| | |
This refactor probably leaves a lot of needless pieces in place; I
tried to do only what was necessary to move away from a raw select()
loop. Nonetheless, I think it simplifies things a bit.
(We can't use asyncio yet since we're keeping compatibility with
older pythons.)
|
| | |
| |
| |
| |
| | |
There are all kinds of possible attribute errors that we would like
to report to the developer.
|
| | |
| |
| |
| |
| |
| |
| |
| | |
Move stable job and update comments.
Add a TODO for 0.4.0 and 0.4.1 nightlies
Closes 30876.
|
| | |
| |
| |
| | |
Part of 30876.
|
| |\ \ |
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | | |
This will help automake's test drivers realize that it has been
skipped.
|
| | | |
| | |
| | |
| | |
| | |
| | | |
Right now a network is unsupported if it requires IPV6 and we don't
have it, if the directory authorities don't actually have dirauth
support, or if one of the binaries is missing.
|
| | | | |
|
| | |/
| |
| |
| |
| | |
We memoize enough things in enough places that we should provide an
actual decorator for this.
|
| | |
| |
| |
| |
| | |
See #23588 for v3 onion service IPv6 in tor.
See #30279 for v3 onion service IPv6 in chutney CI.
|
| | |
| |
| |
| | |
Closes ticket 30820.
|
| | |
| |
| |
| | |
This reverts commit 7782b26767c95988255c5d5884df94580e22c5c3.
|
| | |
| |
| |
| | |
This reverts commit f90f8e4b0d4c8263d0e310a0fa020dda7fe6c91c.
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| | |
Using networks/single-onion-v23-ipv6-md.
Will fail until 23588 is merged into tor master, and built into
a nightly binary.
Closes 30279.
|
| | |
| |
| |
| |
| |
| | |
The previous change only used basic-min with some tor versions.
Part of 30066.
|
| | |
| |
| |
| |
| |
| |
| |
| |
| | |
Some of these networks are aliases for the 0.2.9 networks.
Others test additional features (like HSv3 and consensus IPv6 addresses).
If we do break any 0.2.9 networks, we'll find out in Tor's CI.
Part of 30066.
|
| | | |
|
| | |
| |
| |
| | |
They're not really telling us anything useful.
|
| | |
| |
| |
| |
| |
| |
| | |
Reduces the complexity of the network, decreasing load and increasing
reliability.
Part of 30066.
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
We can't use Linux for IPv6 networks, because Travis Linux does not
support IPv6.
Temporarily disable some unreliable feature tests, until we fix some
bugs.
Part of 30066.
|
| |/
|
|
|
|
|
| |
It's not actually helping us diagnose the errors that we're seeing.
(--allow-failures is pretty good for that.)
Part of 30066.
|
| |\ |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Before these bug 30058 fixes, bootstrap-network.sh:
* Logged the first argument as the network flavour
* default: basic (when there are no arguments)
* Passed $NETWORK_FLAVOUR to chutney as the network flavour
Now, bootstrap-network.sh:
* Uses the first argument as the network flavour
* default: $NETWORK_FLAVOUR (when there are no arguments)
* default: bridges+hs-v2 (when $NETWORK_FLAVOUR is not set)
* This default is the same as test-network.sh's default
* Logs the network flavour
* Passes the network flavour to chutney
Fixes bug 30058; bugfix on commit 822b30c8 in February 2017.
|
| | |
| |
| |
| | |
Part of 30058.
|
| |\ \
| | |
| | |
| | |
| | | |
Merged the shellcheck and unit tests.
Put shellcheck first.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
SC1117 was disabled after shellcheck release 0.5, because it was
too pedantic.
But shellcheck 0.5.0 is still in the Travis Homebrew cache.
So we disable SC1117 in shellcheck-tests.sh.
Part of 30065.
|
| | | |
| | |
| | |
| | | |
Closes 30065.
|
| | | |
| | |
| | |
| | | |
Part of 30065.
|
| | | |
| | |
| | |
| | | |
Part of 30065.
|
| | | |
| | |
| | |
| | | |
Part of 30065.
|
| | | |
| | |
| | |
| | | |
Part of 30065.
|
| | |/
| |
| |
| | |
Part of 30065.
|
| | | |
|