I presented on Thursday at EuroPython. from the feedback at the end it seemed to go well.
I've been told that the video will go up on the EuroPython web site by the end of next week which should be "interesting", as it would be the first time that I've watched myself present, (and I don't do many conference presentations).
What I found different was that I am used to presenting to others within the industry who know about RTL, VHDL, digital simulation, coverage, ... To address this Python-savvy audience I re-wrote a presentation on coverage ranking of Verilog/VHDL and cast it in terms of getting coverage of a Python program and ranking the coverage of three Python tests of the Python program. I then walked them through the ranking algorithm for this much simplified case, then tried to show some of the extra work that needs to be done for the real app.
I decided to not show a more polished version of the script. I showed a version where the main functionality could easily be split into more functions for readability; a psyco import does effectively nothing ... The reason was that what I showed was what I did to convince myself that the algorithm was correct and the speed was fast enough. I would have chucked it at this stage if it was slow. (The need was speed of execution as well as an acceptable algorithm for ranking).
The finish was a "winding down" example of just what you can do when you decide to add documentation for a file format in with the program.
I would like to thank my audience for making someone from an industry they may not have known, feel welcome; and the EuroPython organisers, especially John Pinner, for making it happen.