PyCon double-talk

Tuesday 30 December 2008

It seems a number of people are giving two talks at PyCon 2009 (André Roberge, Jesse Noller, Tarek Ziadé, and Mike Fletcher at least), and I can add my name to that list as well: both of my proposals were accepted.

The first is a straightforward talk about code coverage testing, and coverage.py in particular. I’m hoping to have some major new features in place in time for the talk. Maybe the conference as a deadline will help get it done.

The second talk is more ambitious: how to write a C extension, in 45 minutes. Here, I’ve got a few options how to proceed. One possibility is to go slowly through however much of the topic as will fit in 45 minutes, making sure it is understandable and people come away with a solid footing in the beginnings of the topic. The other way to go is to blaze through as much as I can, firehosing information on the theory that people will be able to go back and read about it later, and it will be familiar to them for having heard it once from me.

Personally I prefer the second style of talk. I learn the general lay of the land, get a quick 10,000-foot overview of the landscape, hear about places to go back to learn more in depth, and so on. But others could feel battered by the rapid-fire barrage of facts with little to grab onto.

Either way, there’s no way to cover even a fraction of the complex topic of writing C extensions. My main goal is to de-mystify extensions so that people coming into the talk feeling like C extensions are deep magic leave the talk feeling like extensions are within the realm of possibility.

I don’t know if previous PyCons had a similar number of double speakers. If not, what does it say about this PyCon?

Comments

[gravatar]
Chad Whitacre 9:00 AM on 30 Dec 2008

+1 for firehose

[gravatar]
Christian Wyglendowski 10:29 AM on 30 Dec 2008

+1 for the firehose. That is my favorite general approach for any of the PyCon talks that I have been to over the past two years. To me, a worthwhile talk is one that I come away from with a better view of the big picture and some tips on where/how to dive into the system/idea/etc that was presented.

[gravatar]
Charles 11:26 AM on 30 Dec 2008

Either is good. The important issue is that you attempt only one. If you choose fire hose, take one, and only one, slide to mention where a workable tutorial is for one to acquire practical experience. If you give a tutorial, then take one slide for the website or map of fire hose topics.

Also, instead of fire hose, think "Map". It's hard to make an explicit framing for all those bits of knowledge, but the frame endures long after the talk.

[gravatar]
John Morrissey 3:27 PM on 30 Dec 2008

+1 for firehose. Talks are most useful to me when they give overviews of a topic, since I can usually sort out the specifics from documentation or experimentation. Overviews are harder to find in docs, plus they benefit from someone's experience.

[gravatar]
Doug Napoleone 8:20 PM on 30 Dec 2008

@Ned:

On the authors presenting multiple talks, we are actually almost at our 07 stats. There are a number of factors skewing things. First we have invited speakers/talks this year for some talks, and at least one 'multiple talk' is due to that. The other is that some people are also on panel discussions. No author is giving more than 2 non-panel discussions. We have more sessions this year in total than in any previous year. Also we had authors proposing many more proposals (two people proposing 5, and many proposing 3 or 4).

In the end the numbers look like this:

A. #authors giving multiple talks (including panels)
B. #authors proposing multiple talks
C. #authors (including panel participants and invited)
D. #proposals (including invited)
E. #talks held (including invited but not keynote)

Year A. B. C. D. E.
2007 10 19 086 104 69
2008 06 24 118 142 64
2009 15 23 091 107 76

It is important to note that we did not have panel discussions last year but did in 07. We did not properly track authors for panels in 07, but did this year. Invited talks are new this year.


So the answer is yes, there are many more people giving more than one talk (50% increase over 2007, almost triple over 2008) but the invited talks, panels, increase in # of proposals per person submitted compared to the number of proposals overall (proposal density), and increased number of slots have all contributed to this.

[gravatar]
Doug Napoleone 8:23 PM on 30 Dec 2008

+1 firehose + openspace followup. Will have laptop w/ compiler on hand :-)

[gravatar]
Doug Napoleone 8:30 PM on 30 Dec 2008

data error...
The number of proposals (including invited and panels) is 118, not 107.
Not that it really matters. I doubt anyone is going to be able to glean any useful information out of this.

[gravatar]
Francesco 2:17 PM on 5 Jan 2009

If I could come (how I wish I could... :-) ) I'd prefer the second style of talk, too [*]. A detailed presentation is better for a tutorial, which you can read and try step by step. But for me it's easier if someone shows me "everything" from the basics to the advanced part, so that I can gain an overall picture and fix the details by myself.
[*] well I hope that some material will be available after the conference...

Add a comment:

Ignore this:
Leave this empty:
Name is required. Either email or web are required. Email won't be displayed and I won't spam you. Your web site won't be indexed by search engines.
Don't put anything here:
Leave this empty:
URLs auto-link and some tags are allowed: <a><b><i><p><br><pre>.