Hi all. After vanishing the last few months I’m delighted to announce a major new Stem feature: type hints!

Python 3’s type information provide benefits familiar to statically typed languages like Java:

IDE autocompletion

pycharm demo

Static type checks

mypy demo

Fresh off the press, this has many rough edges. In particular our library’s internal metaprogramming is incompatible with type data. However, as we refactor our APIs for Stem’s 2.0 release this will only improve.

Aloha! COVID-19 reached Seattle so I’m hunkering down, but prior to that I visited my family in Kona, Hawaii.

Lazy turtles and cute geckos aside, I particularly loved Kona Joe’s roastery tour and Pu`uhonua O Hōnaunau. If you travel to the big island check ’em out!

Family in Hawaii


Deprecation Removal

This month we dropped Python 2 support, and with it 1,200 lines of compatibility hacks. Stem’s next release will require Python 3.6 or greater.

This month I also culled another 2,900 lines of deprecated code. Like any sophisticated library Stem grew organically over the last eight years. Some features are a hit, others clutter our API. Stem 2.x is our opportunity to streamline so expect more rearchitecture going forward.

Winter is such a snoozy season. Eggnog and Stem’s holiday release fading to memory, January I got back to work.

What Big Dreams You Have


Port BridgeDB to Python 3

Python 2’s discontinuation makes it porting season! Stem and Nyx are ahead of the curve, so this month I migrated BridgeDB for Philipp.

This was a lot of work, but only possible thanks to Isis’ great test coverage. My compliments to our former maintainer!


GitHub Migration

Rather than move to GitLab I joined Ooni and part of the Network Team on GitHub. For tickets and pull requests please visit Stem and Nyx on their new home!

Georg asked to subsume responsibility for DocTor so he can move it according to his platform preference instead.

Closing the decade I’m pleased to announce Stem 1.8, the final release in Stem’s 1.x series and with it Python 2.x support.

What is Stem, you ask? For those who aren’t familiar with it Stem is a Python library for interacting with Tor. With it you can script against your relay, descriptor data, or even write applications like Nyx.

https://stem.torproject.org/

So what’s new in this release?


CollecTor Downloader

Through our descriptor archive, CollecTor, Stem can now read Tor’s network topology at any prior point in time.

For example, listing today’s exits is as simple as…

import datetime
import stem.descriptor.collector

yesterday = datetime.datetime.utcnow() - datetime.timedelta(days = 1)
exits = {}

for desc in stem.descriptor.collector.get_server_descriptors(start = yesterday):
  if desc.exit_policy.is_exiting_allowed():
    exits[desc.fingerprint] = desc

print('%i relays published an exiting policy today...\n' % len(exits))

for fingerprint, desc in exits.items():
  print('  %s (%s)' % (desc.nickname, fingerprint))
% python demo.py 
1229 relays published an exiting policy today...

  MrExit (D628F6BB2330B3F78DBB4BED466B0A586D74782E)
  pangea03 (F21DFB7CCD5EEF3E021086EC96EF7CFCAA72F4F3)
  MacarenaValdes (5E3FD31B9DC279C06AD051D68BE08914F6CD3B46)
  TEMPORA (05EAA0696DCB694D6811042348DACD5059FE64AD)
  Quintex43 (1E5136DDC52FAE1219208F0A6BADB0BA62587EE6)

Bandwidth Metrics

Bandwidth Authorities generate the latency heuristics that govern Tor’s path selection. Guiding circuits to be fast, without overburdening individual relays.

With Stem you can peruse this information…

import stem.descriptor.remote
import stem.util.str_tools

bandwidth_file = stem.descriptor.remote.get_bandwidth_file().run()[0]

print('Bandwidth measurements are...\n')

for fingerprint, measurement in bandwidth_file.measurements.items():
  bandwidth = '%s/s' % stem.util.str_tools.size_label(1024 * int(measurement.get('bw', '0')))
  print('  * %s (%s) averaged %s' % (measurement.get('nick', ''), fingerprint, bandwidth))
Bandwidth measurements are...

  * DigiGesTor1e1 (0111BA9B604669E636FFD5B503F382A4B7AD6E80) averaged 23 MB/s
  * WonderWoman42 (E5AA85FA69CDC31900C86E6427C7E5DE11DE9E2D) averaged 37 MB/s
  * alterspalter (B6F0BC2B93CB3EFFFFF724CB4F5E025FB15EFB70) averaged 2 MB/s
  * blueberry (FE80E192AD48A1BEB02D88EBC7663061176E1A79) averaged 1 KB/s
...

Hidden Service v3 Descriptors

With George’s help Stem now reads, decrypts, and even creates HSv3 descriptors. For example…

from stem.descriptor.hidden_service import (
  HiddenServiceDescriptorV3,
  InnerLayer,
  IntroductionPointV3,
)

print(HiddenServiceDescriptorV3.content(
  inner_layer = InnerLayer.create(
    introduction_points = [
      IntroductionPointV3.create('1.1.1.1', 9001),
      IntroductionPointV3.create('2.2.2.2', 9001),
      IntroductionPointV3.create('3.3.3.3', 9001),
    ],
  ),
))
% python demo.py
hs-descriptor 3
descriptor-lifetime 180
descriptor-signing-key-cert
-----BEGIN ED25519 CERT-----
AQgABqvHAX8wXzJY+FqoJQPXNZ8u+SQGPZ1WN/r3hUna0R2AXQnEAQAgBAAuqibl
ALcKa/4nHtLZn2zKV8L4XIpkRyRm7btWPLpYN5Gseb03H5exL+I3SqfG3uNDw5QK
CmPlCQUy3usouSwhO/qWgdy0//bP5kRDma5GDXXWoi3+xTKM6Jez7TGxPAU=
-----END ED25519 CERT-----
revision-counter 1573695064
superencrypted
-----BEGIN MESSAGE-----
aDJodcMjhCvz1K7JCJEAH1H24hvoZ7gZw53AhPdvpHu+5d1Ogwio4qcIXEK1pEgy
QFF1fE6tnCzsk++eMa2WaKwIJYGLPoCnta78H5Ve6VoMj+Pyb5rE6wPTMTPSVm6M
UjllArr7DS8YcofloDxu3iwC3JZYFt/LB6ahq6lBKeot2BD/11pNggkZrZOCLgNQ
pUVyQau7K8ynagVlNNESnI3FccOBaBB4Xa5mObK2ylyiLQ08MqaImW7X2gxeZltT
/C/xtiJXGm2CzkjPpBpMWm09p7/a97GEWca5e8+fhpmGrN7zjAwjYInTvQHS5AyU
7eUFg8ItrRxAiRq4fbe/zepiq2vgfj1Pt7uxC0KCTcLWpd9O/FIvcFSk27Yrtniw
... etc...
-----END MESSAGE-----
signature VDDXXLvgU6qjRI4zfJR3GbQuVjz98qO0LI5gsI60LtGXK2POZ4E+3YVVWuVaEkvMsZaku5qCutIcu74/WQMxCQ

Happy turkey day everyone! George and I were busy with Stem’s HSv3 descriptor branch so I skipped last month’s report, but now have spiffy things to share.

Oh, and wrote a little guide for a mobile game I enjoy and finished unpacking my homely nest!

living room
office


HSv3 Descriptor Support

Stem can now create, parse, and decrypt hidden service v3 descriptors. For example…

from stem.descriptor.hidden_service import (
  HiddenServiceDescriptorV3,
  InnerLayer,
  IntroductionPointV3,
)

print(HiddenServiceDescriptorV3.content(
  inner_layer = InnerLayer.create(
    introduction_points = [
      IntroductionPointV3.create('1.1.1.1', 9001),
      IntroductionPointV3.create('2.2.2.2', 9001),
      IntroductionPointV3.create('3.3.3.3', 9001),
    ],
  ),
))
% python demo.py
hs-descriptor 3
descriptor-lifetime 180
descriptor-signing-key-cert
-----BEGIN ED25519 CERT-----
AQgABqvHAX8wXzJY+FqoJQPXNZ8u+SQGPZ1WN/r3hUna0R2AXQnEAQAgBAAuqibl
ALcKa/4nHtLZn2zKV8L4XIpkRyRm7btWPLpYN5Gseb03H5exL+I3SqfG3uNDw5QK
CmPlCQUy3usouSwhO/qWgdy0//bP5kRDma5GDXXWoi3+xTKM6Jez7TGxPAU=
-----END ED25519 CERT-----
revision-counter 1573695064
superencrypted
-----BEGIN MESSAGE-----
aDJodcMjhCvz1K7JCJEAH1H24hvoZ7gZw53AhPdvpHu+5d1Ogwio4qcIXEK1pEgy
QFF1fE6tnCzsk++eMa2WaKwIJYGLPoCnta78H5Ve6VoMj+Pyb5rE6wPTMTPSVm6M
UjllArr7DS8YcofloDxu3iwC3JZYFt/LB6ahq6lBKeot2BD/11pNggkZrZOCLgNQ
pUVyQau7K8ynagVlNNESnI3FccOBaBB4Xa5mObK2ylyiLQ08MqaImW7X2gxeZltT
/C/xtiJXGm2CzkjPpBpMWm09p7/a97GEWca5e8+fhpmGrN7zjAwjYInTvQHS5AyU
7eUFg8ItrRxAiRq4fbe/zepiq2vgfj1Pt7uxC0KCTcLWpd9O/FIvcFSk27Yrtniw
... etc...
-----END MESSAGE-----
signature VDDXXLvgU6qjRI4zfJR3GbQuVjz98qO0LI5gsI60LtGXK2POZ4E+3YVVWuVaEkvMsZaku5qCutIcu74/WQMxCQ

This branch took quite a lot of work. Many thanks to George and Paul for all the cryptography help!


Other things include…

Hey everyone. This morning my grandma passed away peacefully in her sleep at 6am. She was ninety years old and ready to go. As the saying goes: “old age ain’t for sissies”.

These last few months supporting my family has been a full time job, but my schedule should now free back up.

sunset


CollecTor Support

The only technical thing to note is CollecTor support, a new module within Stem that makes gathering Tor’s network topology at prior points in time simple. For example…

import datetime
import stem.descriptor.collector

yesterday = datetime.datetime.utcnow() - datetime.timedelta(days = 1)

# provide yesterday's exits

for desc in stem.descriptor.collector.get_server_descriptors(start = yesterday):
  if desc.exit_policy.is_exiting_allowed():
    print('  %s (%s)' % (desc.nickname, desc.fingerprint))

Behold, my glorious domain! Weep at its splendor!!1!

granny's attic

This month returned to the island I love, and having a blast volunteering locally. Both non-profits are amazing.

Granny’s is a second hand store that fundraises for our community. Volunteering there is a fun mixture of serendipity with the preeminent form of recycling. I really love matching excess stuff with homes that can make good use of it.

As for our food bank, it feeds fully a seventh of our island’s population. That’s ~150 people! Twice a week I’m unloading trucks and re-bagging groceries. After a decade at Amazon I gotta admit – there’s something deliciously cathartic in non-technical work.

Tor is taking a back seat right now, but I have a feature branch for CollecTor that should be ready to announce soon.

Aside from that lots of runs, plays, and other fun stuff. This is kicking off to be a very fun summer!

Freedom! Sweet, delicious freedom! I went dark since March for crunch time at Amazon, but then decided one decade there was enough. Amazon was a great first job but time for a break.

Vacation aside, this summer I’m volunteering with our island food bank and Granny’s Attic to get more balance away from the keyboard. That said, knowing myself I’ll still need my open source fix.

Oh! And of course, most important of all: found a new puffy friend!

mini puffer

Damn I’m a sucker for puffers.


Tor/Stem CI

Stem’s tests were designed to verify Tor/Stem’s interoperability, as well as provide a tool to test Tor itself. This month teor integrated Stem into Tor’s Travis CI and we’ve been working out the kinks…


Smaller things these last few months include…

Greetings wonderful tor community! This month I pretended to be a web development again for kicks and giggles.


Tor People Page

Hiro’s putting together a new website for Tor. As the maintainer of our people page I ported it over. Personally I think the new page looks a lot better.

Before

people page before

After

people page after

Hrio declined my patch, and that’s fine. Learning Lektor was fun. The new site will launch with Hiro’s version of the page with her as its maintainer.


Personal Website

Beside the Tor website I also put the finishing touches on my own! For comparison my old site is available here.

personal website

Happy January, onion enthusiasts! Made it through another spin of the globe. January I wrapped up a couple projects that were in flight…


Nyx 2.1 Release

My biggest news for users this month is Nyx 2.1! No features, but this release includes a full year’s worth of fixes that smooths Nyx’s rough edges.

If you run into any issues please let me know!


Bandwidth File Support

Bandwidth files are weighting heuristics used by Tor for relay selection. Stem can now read these documents. For example…

import stem.descriptor
import stem.util.str_tools

# Bandwidth files are not yet available, but soon will be...
#
#   https://trac.torproject.org/projects/tor/ticket/21377

with open('/home/atagar/Desktop/bandwidth_file_v1.2') as bandwidth_file:
  desc = next(stem.descriptor.parse_file(bandwidth_file, 'bandwidth-file 1.2'))

print('Median relay bandwidth measurements are...\n')

for fingerprint, measurement in desc.measurements.items():
  # Not quite sure what the units are. The spec seems to indicate KB/s?

  bandwidth = '%s/s' % stem.util.str_tools.size_label(1024 * int(measurement.get('bw_mean', '0')))
  print('  * %s (%s) averaged %s' % (measurement.get('nick', ''), fingerprint, bandwidth))
% python demo.py
Median relay bandwidth measurements are...

  * PompoRelay (319AC0126125DA6A557A416C1BB643E52E272AD0) averaged 187 MB/s
  * oberon (B4486871EA062BAF2354F070FECC344CF3C20880) averaged 196 MB/s
  * ICSIL1xDEDISx1 (7CBBAB7867F939763D94FC6C2819E129971CB0E4) averaged 54 MB/s
  * pollypurebred (EF37D150F3A851B32D9E843605D0EA8A1B938117) averaged 86 MB/s
  * Dementor (46D79CCB83639718177B4BDB8AED6BE9378B1D0F) averaged 158 MB/s
  * wormhole (8B863F5CE8676EFE5BB070CE3CAF54A76AE74D1A) averaged 90 MB/s