Conversation
mfisher87
added
help wanted
|
|
|
Integration tests report: appsharing.space |
|
Interesting, the GitHub build is failing for the same reason as the ReadTheDocs build. EDIT: It was my fault for altering |
|
I am still confused how I managed to break linting 😆 |
mfisher87
force-pushed
the
readthedocs
branch
22 times, most recently
from
ec975ee to
f74e967
Compare
|
Since these lint errors require materially changing the readability of the code failing lint, I decided to disable the rule in this PR. a110a11 Do you all feel we should be forbidding the non-null-assertion operator? cc author of the impacted lines @martinRenou |
Creating multiple environments with mamba exposed lots of problems: * RTD's default machinery expects a single environment YAML file when using conda/mamba. Injecting a 2nd environment into the normal build process gets confusing. * Other hidden assumptions in the default RTD build machinery; I don't remember all the problems I ran in to, but getting a working build required an unsustainable amount of trial-and-error. * Default-installed mamba is too old to support 'emscripten' platform, needs to be upgraded. When upgrading mamba, ran in to a weird issue where the build.html step could not find the python interpreter. It's possible to fully override build steps, but this is an _undocumented feature_. See: readthedocs/readthedocs.org#11551 Instead of overriding build steps, perhaps it will be best long-term to override the entire build with `build.commands`. See: https://blog.readthedocs.com/build-customization/
Makes viewing PR previews more accessible!
This feels significantly more intuitive to me.
|
Lint continued to fail with the same errors in CI on the same rule, even after disabling the rule and passing lint on my local machine. I rebased the PR on |
You need single quotes instead of double quotes in If you run |
|
Thanks all! The lint warnings from eslint were a red herring, and I completely ignored the filenames being output by the prettier check. It's also confusing that I posted in #280 that I'd like to move all the formatting checks into pre-commit to avoid the confusing failures in CI. Thoughts? EDIT: Let's discuss in #281 |
|
Merge #280 first to fix dependabot config |
|
@mfisher87 I merged main to trigger again the check-release test. |
brichet
left a comment
brichet
left a comment
There was a problem hiding this comment.
Thanks @mfisher87, this looks good to me
* [WIP] Attempt RTD build * Try building for RTD in an isolated environment * Restore docs/environment.yml, move docs dependencies to new file * Upgrade the versions of conda and mamba in RTD build * Use Micromamba instead of one of the out-of-the-box tool options Creating multiple environments with mamba exposed lots of problems: * RTD's default machinery expects a single environment YAML file when using conda/mamba. Injecting a 2nd environment into the normal build process gets confusing. * Other hidden assumptions in the default RTD build machinery; I don't remember all the problems I ran in to, but getting a working build required an unsustainable amount of trial-and-error. * Default-installed mamba is too old to support 'emscripten' platform, needs to be upgraded. When upgrading mamba, ran in to a weird issue where the build.html step could not find the python interpreter. It's possible to fully override build steps, but this is an _undocumented feature_. See: readthedocs/readthedocs.org#11551 Instead of overriding build steps, perhaps it will be best long-term to override the entire build with `build.commands`. See: https://blog.readthedocs.com/build-customization/ * Add JupyterLite badge and link to docs * Automate exposing RTD preview link in PR description Makes viewing PR previews more accessible! * Make the RTD PR preview link richer * Ignore the JupyterLite link in the link check automation * Switch bash commands in YAML from folded to literal blocks This feels significantly more intuitive to me. * Format YAML --------- Co-authored-by: Nicolas Brichet <32258950+brichet@users.noreply.github.com>
3 participants
Resolves #208
Hooray! https://jupytergis--275.org.readthedocs.build/en/275/lite/lab/index.html?path=france_hiking.jGIS
Looks like linting is failing unrelated to my change -- perhaps a new release of typescript-eslint modifying detection logic or fixing a bug that was producing false negatives? What's y'all's preference? Deal with the linter errors in this PR or in another PR?
🏗️ Build docs with JupyterLite on ReadTheDocs
💡 Add JupyterLite link to docs index page
🗑️ Remove two bot posts per PR! 🎉 🥳
🦶 New
pr-rtd-link.ymlautomation adds the following footer to each PR:📚 Documentation preview: https://jupytergis--275.org.readthedocs.build/en/275/
💡 JupyterLite preview: https://jupytergis--275.org.readthedocs.build/en/275/lite/