200810 DevSummit EuroBSDCon Strasbourg

Marko gave a short presentation on the current state and further steps: strassburg08.pdf .

• jail commit (may happen at a later point)
• move in constructors intializing globals or variables of “container structures” depending on macro/compile time option (see next item).
• make global vs. “container structures” a compile time option
• bring in destructors, iterators, …
• have a third option (another compile time option) for the indirection →
• with the very next commit after indirection is in, remove the globals, struct.foo option and the compile time options.
• JamieGritton's mgmt/work
• more more more

At one point we will have three different options which code and level of indirection we are going to use.

1. use the globals
2. use the container structures but without indirection (x.foo)
3. use the container structures with the indirection (x→foo)

This will allow us at any time in the future to go back to this revision in the repository and benchmark all three levels with the very same code, should we find a serious regression in the future and want to know, if it was cause by the indirection or not.

GeorgeNevilleNeil was volunteered to help running benchmark tests, especially considering the latency point of view.