Performance of wrapper

[Saelyos]

I'm using pint in my library with some wrappers that give a nice user interface, but unfortunately they also take a lot of time.
Basically I'm using a code that looks like this:

from pint import UnitRegistry

ureg = UnitRegistry()


class Foo:
    def __init__(self, q):
        self.q = q

    @property
    @ureg.wraps("meter", (None,))
    def q(self) -> ureg.Quantity:
        return self._q

    @q.setter
    @ureg.wraps(None, (None, "meter"), strict=False)
    def q(self, value: float) -> None:
        self._q = value

I'm then using self._q (without unit) for some relatively heavy computations.

Unfortunately most of the time is spent in the object creations . This is particularly frustrating with strict=False, when even calling Foo(100) takes a lot of time, as one could expect a more transparent process.

Yet I'm not sure if anything can be done to speed up the process. In case it can help, I've attached the profiler results from my application showing the time spend in pint:

[matroscoe]

I have also been running into significant performance bottle necks as documented in: #1750

[hgrecco]

Performance will be address in the next version. Would you like to help developing or improving the benchmark?

[Saelyos]

I'm not sure how much time I'll be able to spend on this, but I'll try to help improving the performances

[hgrecco]

I pushed some benchmarks. The results are : https://codspeed.io/hgrecco/pint/benchmarks

There are still things to polish, but is a good start.

[hgrecco]

See #1820