Afterward: PyCon on the Charles

The Boston mini-PyCon went well. The Beta House was at max capacity with 30+ in attendance. I met Jesse Noller for the first time; I'm not sure how this managed to be a first because he's a python dev and has been at the last few PyCons. I'd bet there are pictures on flickr that have both of us in frame.

Noller's talk was a thousand foot view of the plethora of multiprocessing/concurrent/messaging/you-name-it frameworks in python. He compared the current proliferation to the mess of competing web and ORM frameworks of two years ago. Sounds about right.

Taylor's talk was about Reinteract, his spreadsheetish interactive python shell. I expected to sleep through this but the app is actually interesting. It falls somewhere between IPython and Resolver One in functionality. I'm sure he and the Resolver guys will have lots to talk about.

My talk went OK. Half the slides are new (again) and I really like the individual slides, but in the rewrite it lost its narrative (the "Radically Simple" of the title). I loved the idea of jumping in on the second slide with an example titled "Why this is Cool" but it fell flat because the first slide didn't explain the pains and foibles of doing without class decorators. I also need to reinsert the longer explanation of what decorators (and metaclasses and mixins) are. The PyCon crowd will be more savvy than the Cambridge user's group but not that much more savvy. A few people said "that looks really cool but I have no idea WTF you are talking about." My talk is flagged as "advanced" in the program but beginner/intermediate/advanced is ignored by attendees (and usually isn't on the printed schedule).

Bruce Eckel will not be giving a keynote as he broke his leg badly while skiing. This is a mixed blessing for me because Bruce's keynote was about class decorators and metaclasses. Good for me because people won't skip my talk for his keynote. Bad for me because I wanted to see his talk and was looking forward to talking to him. Oh, and now I have a useless T-shirt that says "Bruce Eckel Stole My Talk" on the front and "I Stole Bruce Eckel's Talk" on the back.

Ned Bachelder gave his PyCon talk A Whirlwind Excursion through Python C Extensions at the previous meetup. Do click the link, it includes his slides interspersed with his own commentary. I did a similar talk titled "Writing Your Own Python Types in C" a couple years ago so Ned & I traded notes. One of the slides is a nod to our conversation. It isn't as egoboo as the time time Guido gave me full slide w/ attribution in his "State of Python" address, but I'll take it.

PyCon on the Charles

Ned Batchelder has organized a preview/practice session of all the Boston based pythoneer's talks. Ned gave his talk last month at the Cambridge meetup, and March 18th there will be a three hour session with Jesse Noller, Owen Taylor, and me at the Beta House. Talks start at 6:30pm. If you haven't been to a PyCon it is a chance to see a mini version of one. I'll even bring the beer.

Talks are:
Noller: Concurrency and Distributed Computing with Python Today
Taylor: Reinteract: a better way to interact with Python
Diederich: Class Decorators: Radically Simple

[fixed] I misattributed Taylor's talk to someone else.

Python Language Summit

[cross posted from my non-python blog with some minor edits]

I just got my invitation to the Python language summit. I had planned on going anyway, so the invitation just makes it less awkward for everyone involved. The summit overlaps with the tutorial days of PyCon because, well, by definition if you belong in the summit you don't belong in the tutorials [tutorial instructors can suck it].

The summit is interesting because it is unusual* - open source events are usually open access too. The agenda is wide open and is roughly focused on standardization and the future, whatever that is. The invitees are the core developers of all the implementations of Python: regular Python (aka "CPython"), Jython (Java), and IronPython (C#).

I'm not sure what the purpose of the invites is, other than to convey weight. To make the list is to ask "who are the 50 people on the planet who would actually want to come?" Griefers and trolls would be bounced regardless [come to think of it, maybe I was invited because I have no compunctions about bouncing griefers and trolls] so perhaps the invitations are meant to discourage the well meaning but clueless. Every convention has at least a few of those - does anyone remember the "callable None" guy from a few years back? "well meaning but clueless" doesn't begin to describe him.

* It is not without precedent. See the the Reykjavik sprint. [my favorite bit from those posts "Cod jerky smells a bit like feet but tastes OK. Dried shark smells a lot like feet and tastes exactly like dried asshole."]

PSA: Wireless Keyboards

The batteries in your wireless keyboard don't die at arbitrarily long intervals. You put your cellphone close to the receiver at arbitrary intervals and then move it while changing the batteries.

And by "you" I mean "me."

Look Who's [not] Talking at PyCon

The schedule is up and it looks good. It is a mix of conference warriors plus some new blood; likewise the talks are a mix of old standards and new topics (some of the new talks are by conference warriors and vice versa).

The familiar names include Brett Cannon (cpython), Jim Baker, (jython), and Michael Foord (ironpython). There is a host of names that might be missing or might not - I can't recall if they do talks every year - but there is no Norwitz, Warsaw, Holden, or Martelli. Noticeably absent is Raymond Hettinger who has given several talks per con at several cons a year. Noticeably present is EVE Online in the form of Richard Tew and Krisjan; with the Icelandic Krona where it is I don't know how they can afford a taxi let alone air fare.

On the new list is Bill Gribble. Bill Gribble gets a free beer for having one of the best names ever (if his Mother shows up she can collect for her work, instead). I bet none of his friends ever start a story "I was having lunch with my friend Bill..." But instead always "I was having lunch with my friend Bill Gribble..." I don't know Gribble from a hole in the wall but I'll make a point of changing that.

[update] Talked to Raymond and he'll be attending, at least (barring life's usual caveats).
[updateder] As Brett and Ivan mention in the comments there is now a tier of invited talks that includes many of the "missing" conference regulars. Glad to have em, the "hall track" is my favorite and many on the invited speakers list make it so.

Post's Machine

An undergrad CompSci major, Shriphani Palakodety, posted an implementation of Post's Machine. Post's machine is a simple computer that has separate data and execution storage. Here's my implementation and my advice to Shriphani - Keep it simple! When working on hard problems post-graduation you'll find that code gets complicated of its own volition. The job of the writer is to fight it at every turn (the next guy to come along will appreciate the effort).

This version replaces the two custom containers with python dicts and moves all the complicated code to the pretty printer. A collections.defaultdict might be a slightly better choice for the 'infinite tape' that starts as all zeros. A nice thing about defaultdicts is that you can min() and max() even empty dicts.

I also did a python3.0 version which was nearly identical except for a 'with' on the file open and print-as-fucntion. Disappointingly I thought the advanced tuple unpack syntax would help in the case where a too-short tuple is padded and then the padding discarded.

# args might be a two or three tuple
a, b, c = (args + [0])[:3]

# python3 syntax
a, b, c, *ignore = args + [0]

The *ignore argument demands to be read as opposed to the [:3] trimmer on the end which keeps the low profile it deserves.

Here's the code for my Post Machine

def parse(lines):
  program = {}
  for line in lines:
    pos, action, jump = [p.strip() for p in line.split(',') + ['0']][:3]
    program[pos] = action, jump
  return program

def execute(program):
  tape = {}
  tape_pos = 0
  action = None
  action_pos = '0'

  while action != 'exit':
    action, action_pos = program[action_pos]

    pretty_tape(tape, tape_pos)

    if action == '<':
      tape_pos -= 1
    elif action == '>':
      tape_pos += 1
    elif action == 'mark':
      tape[tape_pos] = 1
    elif action == 'unmark':
      tape[tape_pos] = 0

  return tape, tape_pos

def pretty_tape(tape, tape_pos):
  if not tape:
    tape = {0:0}
  min_pos = min(tape_pos, min(tape), 0)
  max_pos = max(tape_pos, max(tape), 4)

  parts = []
  for pos in range(min_pos, max_pos + 1):
    val = tape.get(pos, 0)
    if pos == tape_pos:
      parts.append('[%d]' % val)
    else:
      parts.append('%d' % val)

  print '.....', ', '.join(parts), '.....'

if __name__ == '__main__':
  program = parse(open('post.txt'))
  execute(program)

PyCon Talk Accepted

I received my talk acceptance email yesterday. On schedule, even! The PyCon organzers get more polished every year (I didn't submit a talk last year so maybe this isn't new). It will be interesting when they publish the stats for talk proposals. I vaguely recall that the rate was 50% last year and 80% the year before that. The attendee numbers doubled between '06 and '07 so the number of proposals likely did too.

The email included the anonymous reviews of the proposal. Most were +1 but I'll keep in mind the +0 and -1s when preparing the new version of the talk. They were mostly concerned that the PyCon UK version of the slides was too code heavy. There is even more material now that decorators are in use "in the wild" so I may shunt those into a separate tutorial-like paper or maybe just include them as slides after the end so people who download the .pdf will see them.

I'll definitely have to prep and practice the new talk well in advance of PyCon. Lots of possibilities there; The Colorado Python group get together to practice all their member's talks every year - we could get the Boston PIG to do something similar. And I could make a show-me-do thing. And I'll definitely give it to the pythoners at my old company as a trial run.

Don't forget to register early for PyCon. We maxed out the hotel some nights last year and this year should be even bigger.