-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[ci] [docs] use miniforge for readthedocs builds (fixes #4954) #4957
Conversation
Sure, done! |
Thanks for that! I'm really happy to see that the build succeeded, docs look correct (I checked Python, C API, and the R docs), and that the build didn't take longer than usual. |
Hmm...
https://github.com/microsoft/LightGBM/runs/4863199597?check_suite_focus=true |
Co-authored-by: Nikita Titov <[email protected]>
I think I understand what's happening. I've found reports claiming that this error is about a difference in
The It looks like try: # Python < 3.10 (backport)
from importlib_metadata import entry_points
except ImportError:
from importlib.metadata import entry_points I suspect that the issue here is related to mixing conda channels. Notice in https://github.com/microsoft/LightGBM/runs/4863199597?check_suite_focus=true that
I think that specifying
I tried that approach because |
I haven't been able to figure out the right combination of Since the builds are working correctly on RTD (build link) and in Docker locally (both using So I'd like to request changing the order of merging PRs I'd originally suggested in this PR's description. @StrikerRUS could we merge #4953 before this one? Then I'll update this one to use |
Sure! |
Now that #4953 has been merged, I'll work on updating this. https://docs.readthedocs.io/en/stable/config-file/v2.html#build-tools-python says it's possible to set |
Ok I think this is ready for review!
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you so much for working on this! Please consider checking my fresh comments below.
- r-base=4.1.2 | ||
- r-data.table=1.14.2 | ||
- r-jsonlite=1.7.2 | ||
- r-knitr=1.37 | ||
- r-matrix=1.4_0 | ||
- r-pkgdown=1.6.1 | ||
- r-rmarkdown=2.11 | ||
- r-roxygen2=7.1.2 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We don't want to pin R package versions anymore?.. #2176 (comment)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The comment you've linked refers to a commit, 014bee1, that changed the list of package from "not pinned at all" to "pin to aa specific BUILD of a specific VERSION".
This PR's changes are still pinning them to specific package VERSIONS, just not to individual BUILDS. Sometimes new builds of the same package version are pushed to conda channels, to fix issues like, for example, mistakes in metadata about declared dependencies. We should want to get such fixes automatically, as they make it more likely that the environment will solve correctly.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The comment you've linked refers to a commit, 014bee1, that changed the list of package from "not pinned at all" to "pin to aa specific BUILD of a specific VERSION".
More relevant past discussion: #2725 (comment).
OK, let's not pin build numbers for now. But if RTD starts to fail, we know what to try first.
Co-authored-by: Nikita Titov <[email protected]>
… into fix/rtd-configuration
- r-base=4.1.2 | ||
- r-data.table=1.14.2 | ||
- r-jsonlite=1.7.2 | ||
- r-knitr=1.37 | ||
- r-matrix=1.4_0 | ||
- r-pkgdown=1.6.1 | ||
- r-rmarkdown=2.11 | ||
- r-roxygen2=7.1.2 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The comment you've linked refers to a commit, 014bee1, that changed the list of package from "not pinned at all" to "pin to aa specific BUILD of a specific VERSION".
More relevant past discussion: #2725 (comment).
OK, let's not pin build numbers for now. But if RTD starts to fail, we know what to try first.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks a lot! Just two minor suggestions below and I think this PR can be merged.
Co-authored-by: Nikita Titov <[email protected]>
This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this. |
Fixes #4954.
Contributes to #4948.
Contributes to #4859.
This PR proposes updating LightGBM's readthedocs configuration to tell RTD to use
mamba
to manage dependencies. See the description of #4954 for relevant links to RTD documentation describing these new config options.Notes for Reviewers
This PR updates the
"check-docs"
CI job to useconda
. I'm hoping this can be merged before #4953, and then #4953 could switch that CI check to usemamba
so LightGBM's CI will more closely match what happens onmaster
RTD builds.@StrikerRUS could you please temporarily enable builds for this branch on RTD so we can test the new configuration?