Repository navigation
Fix sphinx automation - #162
Conversation
|
@petercorke , it appears that the sphinx documentation workflow was dependent on a repo you had forked, |
|
I made some mods to speed it up, and change it to use project.toml, but that file is broken. Will look at that next |
| <img src="https://github.com/bdaiinstitute/spatialmath-python/raw/master/docs/figs/CartesianSnakes_LogoW.png" width="200"></td> | ||
| <td style="border:0px"> | ||
| A Python implementation of the <a href="https://github.com/petercorke/spatial-math">Spatial Math Toolbox for MATLAB<sup>®</sup></a> | ||
| A Python implementation of the <a href="https://github.com/bdaiinstitute/spatial-math">Spatial Math Toolbox for MATLAB<sup>®</sup></a> |
There was a problem hiding this comment.
revert, it refers to the MATLAB ancestor which is surely not at BDAI.
There was a problem hiding this comment.
Ah, sorry about that!
petercorke
left a comment
petercorke
left a comment
There was a problem hiding this comment.
I believe I have fixed the sphinx-auto problem, bad pyproject.toml file. It is probably a 100x faster now, but just go with what's most expedient for you. It has revealed quite a few execution errors in the example code blocks (the example blocks are wrong, not the toolbox), I will make a PR for these, but this shouldn't stop us publishing.
There are a few incorrect changes from petercorke to bdaiinstitute when referencing robotics toolbox, RVC book, MATLAB ancestor of this toolbox.
A lot of changes seem to be just trailing white space.
Sorry about that, I did a blanket replace of "github.com/petercorke" as a stab at figuring out what was going on with the sphinx stuff and thought I had reverted it already. |
…hub.com/bdaiinstitute"
I thought the black formatting PR I just merged would have also fixed this for all the files as well. I'll probably make one for that then. |
petercorke
left a comment
petercorke
left a comment
There was a problem hiding this comment.
This looks great, thanks
Resolves conflict in .github/workflows/sphinx.yml: keep master's
dynamic ${{ github.repository }} reference for the gh-pages clone
step instead of this branch's stale hardcoded bdaiinstitute org name
(pre-dates the org rename to rai-opensource).
…ps (#221) ## Summary `sphinx-pyrunblock` is Peter's own package (a hard fork of the original `WhyNotHugo/sphinx-autorun`, published properly to PyPI under this name). Today's CI installs it via `pip install git+https://github.com/petercorke/sphinx-autorun.git` -- an old, pre-rename URL that still works only because GitHub redirects renamed repos. This switches to installing the plain PyPI wheel instead. - Bumps `pyproject.toml`'s `docs` extra from `"sphinx-autorun"` to `"sphinx-pyrunblock"` (the old entry was misleading in isolation -- without the workflow's follow-up git install, it would resolve to the unrelated upstream PyPI package of the same name, not Peter's fork). - Drops the now-redundant `pip install git+...` line from `sphinx.yml`, since `pip install .[dev,docs]` already installs the right package. - Updates `conf.py`'s `extensions` list from `sphinx_autorun` to `sphinx_pyrunblock` to match. ## No functional change Both the old git URL and the new PyPI package resolve to the identical `sphinx-pyrunblock==1.1.0` release today -- confirmed with a fresh, uncached `pip install git+...` and comparing installed package metadata. This is purely a speed/hygiene change. ## Why The PyPI install is consistently 2-4x faster in a clean venv (no git clone/VCS metadata resolution): | Method | Run 1 | Run 2 | |---|---|---| | `git+https://...sphinx-autorun.git` | 20.5s | 10.6s | | `sphinx-pyrunblock` (PyPI) | 5.4s | 7.1s | That's ~9-15s saved on every docs-build CI run (every push to `master` plus every PR). ## Test plan - [x] Fresh isolated venv, `pip install ".[dev,docs]"`, confirmed `sphinx-pyrunblock==1.1.0` resolves without the git install line - [x] Real `sphinx-build -b html` against the rebased branch (includes #198, #162) -- succeeds, 15 warnings, consistent with pre-change baseline Co-authored-by: Tim Perkins <code@taughz.dev>
2 participants
The workflow to generate sphinx documentation seems to not be working. Trying to fix that with this PR.