- Nov 20, 2016
-
-
Damian Johnson authored
Release creation in PyPI is both error prone and unfixable. As such it's safest if we have a scrap project to sanity check that things are working as intended before permanently screwing up our package. Really silly we need to do this, but that's life. https://github.com/pypa/packaging-problems/issues/74#issuecomment-260716129 Previously we made releases with 'python setup.py upload' but PyPI has broken this by prohibiting non-SSL uploads. Twine now seems to be the only game in town so updating our setup.py clauses with things we previously set via the site. One caveat is that PyPI's bug tracker field evidently cannot be set via the setup.py... https://stackoverflow.com/questions/14459828/how-to-set-bug-tracker-url-in-setup-py-script Oh well. Close enough.
-
- Nov 17, 2016
-
-
Damian Johnson authored
Damnit PyPI, you cannot legitimately claim this isn't a bug. Long story short if a file is improperly uploaded that version is left in an unfixable state, forcing us to do a version bump. https://github.com/pypa/packaging-problems/issues/74#issuecomment-260716129 No changes to Stem here. Just bumping the version to get around their bug.
-
- Nov 14, 2016
-
-
Damian Johnson authored
-
Damian Johnson authored
Usually I wouldn't bother for a minor one-off change but we're about to cut a release, so might as well make sure we're fully up to date.
-
- Nov 12, 2016
-
-
Damian Johnson authored
One trouble with the ONLINE testing target is that it's run so infrequently. Tor manual changes are causing a few assertion failures. Correcting them.
-
- Nov 06, 2016
-
-
Damian Johnson authored
On #17306 neel's test output has quite a few pyflakes issues I don't get... https://trac.torproject.org/projects/tor/attachment/ticket/17306/integ_test.txt These would be a hassle to suppress one by one so allowing our config to suppress by prefix or suffix.
-
-
Ephemeral hidden services do not touch disk so there is no hidden service directory to display.
-
-
The xrange function no longer exists in Python 3 [0]. [0] https://docs.python.org/3.0/whatsnew/3.0.html#views-and-iterators-instead-of-lists
-
-
Damian Johnson authored
Vidalia link no longer worked and trac components are now nested.
-
Damian Johnson authored
Simple error with our new test... ====================================================================== FAIL: test_installs_all_data_files ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/atagar/Desktop/stem/test/unit/installation.py", line 82, in test_installs_all_data_files self.fail("setup.py doesn't install %s" % path) AssertionError: setup.py doesn't install /home/atagar/Desktop/stem/stem/.settings.cfg.swp ---------------------------------------------------------------------- Ran 2 tests in 0.018s FAILED (failures=1)
-
Damian Johnson authored
As tor adds new config options we want short, particularly human-readable summaries. Filling in the new additons. We have a test that checks for this but it required the ONLINE integ target so really never gets run in practice. Replacing this with a unit test since honestly we don't need the downloaded information for this one.
-
- Nov 04, 2016
-
-
-
-
Damian Johnson authored
Oops, good catch on... https://trac.torproject.org/projects/tor/ticket/20477#comment:5
-
Damian Johnson authored
Another easy gotcha that's bitten me before. Like modules, non-source files need to be explicitely listed.
-
Damian Johnson authored
I've screwed up on this enough times that we might as well have a test for it. :P
-
Damian Johnson authored
Oops, missed a couple spots where we still call it pep8.
-
- Nov 03, 2016
-
-
Damian Johnson authored
Interesting, I use vim but have never seen these. Maybe it's something with newer versions? Regardless, great catch from Chelsea... https://trac.torproject.org/projects/tor/ticket/20520
-
Damian Johnson authored
Minorest of minor nitpicks. Was perfectly fine, just prefer a blank line and for exclude to be in the conditional (imho didn't previously add any readability).
-
required files
-
-
- Nov 02, 2016
-
-
Damian Johnson authored
Fantastic, turns out tor supports this! https://trac.torproject.org/projects/tor/ticket/20501#comment:5 Adding a new get_consensus() arg for getting the microdescriptor consensus.
-
- Nov 01, 2016
-
-
Damian Johnson authored
Just a few small tor changes since we last synced.
-
Damian Johnson authored
Updating our fallback directories caused a big delta, not because of the changes but because we have a random ordering. Sorting our config file by the fallback fingerprints so we have small deltas in the future. Only functional change to the config file is the following to get us in sync... * Removed FCB6695F8F2DC240E974510A4B3A0F2B12AB5B64 as a fallback directory * Removed 14419131033443AE6E21DA82B0D307F7CAE42BDB as a fallback directory * Removed 5525D0429BFE5DC4F1B0E9DE47A4CFA169661E33 as a fallback directory * Removed 29F1020B94BE25E6BE1AD13E93CE19D2131B487C as a fallback directory * Removed E781F4EC69671B3F1864AE2753E0890351506329 as a fallback directory * Removed DE134FC8E5CC4EC8A5DE66934E70AC9D70267197 as a fallback directory * Removed 6A640018EABF3DA9BAD9321AA37C2C87BBE1F907 as a fallback directory * Removed 6FC6F08270D565BE89B7C819DD8E2D487397C073 as a fallback directory
-
Damian Johnson authored
This was a great test at one point but tor's added so many fallbacks that it takes *forever*. At a rough guess maybe a half hour. Better as a DocTor check anyway (think we actually have one already...).
-
- Oct 30, 2016
-
-
Damian Johnson authored
Interesting. Teor ran into an issue due to lsof returning three (rather than the expected two) values on OSX... https://trac.torproject.org/projects/tor/ticket/20477 Not quite sure why but meh, easy to fix.
-
- Oct 29, 2016
-
-
Damian Johnson authored
Good point from Tom that our docs around this are wrong... https://trac.torproject.org/projects/tor/ticket/20493 https://gitweb.torproject.org/torspec.git/tree/dir-spec.txt#n2012
-
Damian Johnson authored
The pep8 project renamed itself to pycodestyle at Guido's request... https://pypi.python.org/pypi/pycodestyle Supporting both the old and new name.
-
Damian Johnson authored
Supporting the old pep8 function and config as an alias. Minor version bumps shouldn't break users. ;)
-
Damian Johnson authored
Just a couple missed spots.
-
Damian Johnson authored
Lets not vend this helper to Stem users just yet. Also fixing pycodestyle regressions that were introduced as part of the prior changes. :P
-
Damian Johnson authored
Reasserting our testing configuration of pycodestyle issues to ignore.
-
-
-
Damian Johnson authored
When expanding the 'private' alias for exit policies the last entry is for our pubic address. When python is unable to determine one get_config_policy() excludes this last entry, in turn causing our unit tests to fail... https://trac.torproject.org/projects/tor/ticket/20476 Being more careful to treat this part of the policy as being optional.
-
- Oct 27, 2016
-
-
Damian Johnson authored
get_config_policy() accepts an optional address for expanding the 'private' alias. Trouble was that with IPv6 addresses it wouldn't accept an address with brackets, but providing an address without brackets in turn was rejected by ExitPolicy class validation. Accepting both.
-
- Sep 20, 2016
-
-
Damian Johnson authored
When running with the python 2.x if you have the queue module we use io.StringIO rather than StringIO.StringIO. This causes two of our Controller's methods to fail with... Traceback (most recent call last): File "circuits.py", line 7, in <module> for circ in sorted(controller.get_circuits()): File "/usr/lib/python2.7/site-packages/stem/control.py", line 455, in wrapped return func(self, *args, **kwargs) File "/usr/lib/python2.7/site-packages/stem/control.py", line 3284, in get_circuits circ_message = stem.socket.recv_message(StringIO('650 CIRC ' + circ + '\r\n')) TypeError: initial_value must be unicode or None, not str Thanks to mancha for catching this!
-