wurst code profiler

[Cokemonkey11]

Is your feature request related to a problem? Please describe.

Hive~fags~rs often have a "perf-first" approach to writing jass. Sometimes it's justified since jass is so slow. Other times, it's silly and leads to shitty, inlined, repetitive code.

A profiler can help justify not writing shitty code. "Measure, then improve" is a much more powerful argument when we have measurement tools.

Describe the solution you'd like

• we could ask Dracolich for cost model for every jass native and basic operation
• build a mapping function
• and then every wurstunit test could calculate cost and draw a perf-style graph
• or maybe a callgrind-style hot path backtrace

[Frotty]

A profiler can help justify not writing shitty code

I'm not quite sure how. It might prove a point, but people won't change their coding habits just from seeing some profiler results imo. It could even lead to the opposite happening - people trying obscure things to get better profiling results. Premature optimization is bad and that should be embraced. People need to notice the benefits of not writing "shitty" in action and not by some measurements.

Stdlib docs, tutorials, etc. would be easier to implement and more beneficial imo.

[Cokemonkey11]

Stdlib docs, tutorials, etc. would be easier to implement and more beneficial imo.

I agree, and on that basis I'm pretty happy to call this not high prio