Peter Hill

I've discovered that the vast majority of the warnings emitted from Sphinx are due to two separate bugs in doxygen and breathe. Things that still need doing: - [ ]...

The tests are complaining about some problem in the adams bashforth solver, which I don't think was changed here -- possibly bad merge?

Bah, there is a C++17DR for making `std::visit` work when inheriting from `std::variant`. The fix is in gcc 11 and clang 15. This works fine in Intel. So options are:...

@dschwoerer The aim of this work is to ensure that `Coordinates` is always correct, and that modifications to the metric are correctly and automatically propagated to the derived quantities, as...

> If we do add wrappers, should we not add them now? I want to make sure we get the features in this PR correct first, and then look at...

Sorry @johnomotani, I missed this at the time. - Good spot, it should be `netCDF_ROOT`. I don't particularly like this weird mix, but that's how the netCDF team style it,...

I also run into this a lot, and I just enabled `toggle-debug-on-quit` and hammered `C-g`, which gave me the following stack trace: ``` Debugger entered--Lisp error: (quit) re-search-forward("\\s|" nil t)...

It looks like [`conda.cli.common.specs_from_url`](https://github.com/conda/conda/blob/239ce89fbd17a560cf49d9297a67bd48afa8467d/conda/cli/common.py#L153) handles parsing `requirements.txt`, so is it just a matter of something like: ```python def specs_from_url(url, json=False): from ..gateways.connection.download import TmpDownload explicit = False with TmpDownload(url, verbose=False)...

The above patch does provide some basic support for `pyproject.toml`, but after some quick testing on conda's own `pyproject.toml`, `spec_from_line` doesn't fully support [PEP 508](https://peps.python.org/pep-0508/), failing on `"menuinst >=1.4.11,

We've just been bitten by this. What about the possibility of restoring the `error stop` as the simplest implementation that ensures test failures have non-zero exit code, and include in...