Closed TheTripleV closed 10 months ago
Maybe you use a dated version. Tests cover this.
Ok, I forgot about case-sensitivity. :disappointed:
CI tests against pybind11's master
branch, which has capitalized Buffer
in docstrings.
I should probably include older pybind11
version to CI :(
That's weird, because I updated our pybind11 recently (but we use smart_holder)... which has b4573674bc65a7842b49be45d63b582f4c15d1b4 in it?
commit 7953d19a7c17ff1bfe0b7c4cdfde216d400d5f28 (HEAD -> 2024, origin/smart_holder)
Merge: 9110b766 2b2e4ca4
Author: Ralf W. Grosse-Kunstleve <rwgk@google.com>
Date: Tue Oct 3 22:45:09 2023 -0700
Merge branch 'master' into sh_merge_master
...
commit b4573674bc65a7842b49be45d63b582f4c15d1b4
Author: Sergei Izmailov <sergei.a.izmailov@gmail.com>
Date: Wed Sep 13 04:47:39 2023 +0900
Update render for buffer sequence and handle (#4831)
I see 2.3.4 failing at https://github.com/robotpy/robotpy-wpiutil/actions/runs/6609220896/job/17949256791#step:8:547
Ah, our span_caster outputs a lowercase buffer (https://github.com/robotpy/robotpy-wpiutil/blob/2dd9b691b7dc776fd266f04f4b83bef45dfdef23/wpiutil/src/type_casters/wpi_span_type_caster.h#L148)... but I suppose you should support either the older or newer version.
This is actually the third type caster related issue that I've ran into where the name was deceiving me. I don't think there's a way for stubgen to know if the type has a custom caster associated with it -- if it did, that would be really useful to output in the error message.
I don't think there's a way for stubgen
That's correct, stubgen is unaware of casters
Thank you all for the issue and quick response :+1:
CI tests against pybind11's master branch, which has capitalized Buffer in docstrings. I should probably include older pybind11 version to CI :(
@sizmailov Do you think it is easy to test in CI / support in stubgen the latest stable pybind11 release (besides master
)? :)
@ax3l I can add v2.11
branch to CI if it's what you are asking for.
Thank you, that sounds great and was what I was thinking.
buffer
now seems to look fortyping.Buffer
(which doesn't exist) and nottyping_extensions.Buffer
.