Caterpillar

Some time ago, I talked about stepping backwards in the debugger, and how event sourcing could play a role in that. I also floated the idea of a distinction between solid code, that is highly optimized; and fluid code, which provides live interaction.

Now I'm wondering, can we combine both ideas? By default, you can step through your code forward and backward, expression by expression. If you have a piece of code that you need to run faster, you can tell the development environment to optimize it, losing some interactivity.

But what if your code used event sourcing and the system understood how that worked? Then this distinction between optimized "solid" and interactive "fluid" code could be made along the boundaries set by the events. You could still step backwards, just in a more coarse-grained manner; event by event.