Again and again, we talk about the performance impact of a change. But it is quite often really hard to benchmark the effect of a change. I propose we should include benchmark information on every pageload, or someting like that, coupled with an analysis (integrated in the backend?) of what takes how long, and what takes now longer than before, and how long it should take.
Maybe - as a start - we can include xhprof for something like that, it impressed me quite a bit in https://github.com/s9y/Serendipity/comm ... nt-6342952. And maybe we can find some helpful visualisations, instead of just showing the numbers.
If something like this is enabled by default when in debug mode (or rather: not in production), it could guide future developments/developers to be better aware of the performance situation.
This is future, not 2.0 (ofc, if someone has a solution on hand that might change)