[v4] Memory usage for usertypes (especially containers) is excessive

[ThePhD]

There's a number of scenarios where sol2 is making garbage that it seems not to be destroying, as evidenced by a few issues filed recently about leaks (particularly, on *Nix platforms such as #1238 and #1094 and #1256).

This issue will be for driving down the both binary and runtime usage of sol2 into the goddamn ground.

• [ ] Drive binary sizes down into the ground (goal: cut all tests compiled binary sizes in half without -O0, reduce object file size by at least 40%).
• [ ] Reduce string serialization usage down dramatically (source: class names are a constant source of differing string constants and should be slimmed down excessively using a prefix-based optimization scheme, by hand to guarantee it).
• [ ] Find out what is causing excessive kept-alive memory using sol2 with functions and similar when looping over code (use test code from #1256 or #1238).

[Domain]

Any update? Memory leak is very painful.

[Nrosa01]

I'm too interested in this, this makes it hard to use for a game engine

[ThePhD]

No updates so far, only the promise that I'll get to this once my life calms the hell down. I would say "pull requests welcome" but even I know the internals are a soupy mess (which is why I want to rewrite them, to enable better contributions from others).

But getting the activation energy and free time to kick this off is, uh.

Somethin'.

[Codinablack]

No updates so far, only the promise that I'll get to this once my life calms the hell down. I would say "pull requests welcome" but even I know the internals are a soupy mess (which is why I want to rewrite them, to enable better contributions from others).

But getting the activation energy and free time to kick this off is, uh.

Somethin'.

Sounds great! You do what you got to do. I think we all can relate to not having enough time, life is busy. Thanks again for all your hard work!