Because of time constraints I didn't present any talk at Singapore PyCon, but did step up and do one in Sydney which I then gave again at New Zealand. I didn't do anything for US PyCon as I only found out I would be going a bit more than a month out from the conference.
The conference talk I did last year was one on getting started with mod_wsgi. The approach the talk took was to show all the things that could go wrong and explain why those situations come about. Having now attended the US PyCon, in hindsight I probably tried to pack a bit too much into the talk and one could argue it was also perhaps too instructional for a conference talk. Most talks seem to be much more high level and along the lines of 'hey look at what I am doing' rather that being a mini tutorial intended to actually teach you a lot of stuff while you are there.
For that reason, rather than try and do further highly technical talks on WSGI deployment or debugging of web applications, I thought for this year that I would instead step back a bit and do a talk on the basics of how web servers work. I realise that this isn't necessarily Python specific, but what keeps coming up out there on the various forums is that many people simply don't understand how the web server they are using actually works. Most see the web server as merely an inconvenience, something that they have to install but which they never understand enough to know how to set it up properly.
The talk I have in mind would therefore cover the common architectures that web servers use. So, topics such as threaded vs event driven systems, single process vs multi process, as well as descriptions of the different request queuing mechanisms and use of backend daemon processes for running the dynamic parts of a web application. There obviously needs to be some Python slant to all this, so would also cover how the architecture of the web server affects the type of interface used between a Python web application and the web server, plus issues such as data sharing between different parts of the Python web application. Finally, would try and cover why you would choose one server architecture over another or when you should actually employ multiple web servers with different architectures for different parts of one overall Python web application.
So my question is, is a talk about such a topic something that would be of interest out there in the Python web community? I sort of feel it is an area where there isn't a great deal of good information out there about, but I could be totally wrong.
If not the above topic then, what would people like to hear me do a talk on? Realistically I might only have time to prepare one good talk, but if there is interest in other topics, I can always try for more. I do have to work it out soon though to get in before call for papers for Singapore and Sydney PyCon conferences closes.