WIP: Initial rust build support.

[pelme]

This is just ensure that tests can invoke the built rust code to make
sure that the test/build tooling works.

[choucavalier]

what is the goal of using rust in htpy? (just curious!)

[pelme]

improved performance! played around with it but it seems to rather make performance worse since there are a lot of small/cheap calls happening. #157 is likely a better approach for improved performance. I have started working on a benchmarking suite to actually understand the performance of htpy (and templates and other HTML generation tools) better.

[choucavalier]

i see! an interesting approach is compile time templates by Maud. probably out of scope but still interesting

[pelme]

Maud looks like something I would use to generate HTML in Rust. 🙂

I think ast transforms is as close to compile time templates we can get here in Python. :)

My current plan is roughly:

• Implement a decent suite of benchmarks to actually know what improves/degrades performance. Compare against jinja as a baseline for what decent performance looks like.
• Implement ast transforms that transforms calls to article("#featured.bold")[p[h1["Hi"], text]] to something like CompiledElement('<article id="featured><p><h1>Hi</h1>' + escape(text) + '</article>'). Implement CompiledElement as Python.
• Investigate if implementing CompiledElement in Rust would improve performance even more.