Gustavo Nunes Goretkin

I missed the `convert` function. Thanks! Suppose I want to avoid having to choose a name like `rational_to_number` (in example from documentation, copied below), especially because I will not plan...

@rtbs-dev thanks for pointing this out! I gave it a try [here](https://github.com/ggoretkin-bdai/investigate-plum-performance/commit/fbac9b8e96fdb91a7f1778a1ff35e7545be8cedd) and it seems to work. Unfortunately, the performance is worse than using `plum.convert`, but I have not dug...

Profiling reveals that `is_bearable (beartype/door/_doorcheck.py:317)` is the hot spot.  I believe it would be ideal for `plum.convert` to be implemented in terms of `typing.Type`, if the performance can match....

> @ggoretkin-bdai Just so I'm clear, you are trying to add a dispatch rule to the base `convert` method (which is supplied by plum) and want to specify it via...

> Currently, caching happens based on the type of the arguments. Perhaps we could extend this hashing to values. I.e., if `f(x)` is called, currently caching is based on `type(x)`,...

I want to check if we're aligned in the high-level philosophy here, so I'm going to try to explain my understanding of the situation. It's clear to me that `plum`...

From searching the code, it seems that `--stacks` has no effect.

It seems like the `l4t` images are missing these files: ``` $ docker run --platform linux/aarch64 --rm -it nvcr.io/nvidia/l4t-base:r32.5.0 find / -name "libcublas*" 2>&1 | grep -v "Permission denied" $...

ah, sorry, the last message is not quite relevant. This is the valid test (need the nvidia runtime) ``` $ sudo docker run --runtime=nvidia --platform linux/aarch64 --rm -it nvcr.io/nvidia/l4t-base:r32.7.1 find...

From https://github.com/dusty-nv/jetson-containers/issues/205#issuecomment-1310393417 it seems that the intention is not that this image can be built on `x86`. Building it on the Jetson means that the default docker runtime is nvidia,...