[ci] use GitHub Actions to re-generate R configure #4140
Conversation
There was a problem hiding this comment.
because anyway we need a follow-up commit with reverting back to
VERSIONin configure.ac, right?
No, no followup commit is required because there is never a commit that removes the ~~VERSION~~ in configure.ac. I'm confused, how is configure.ac getting updated and showing up in the diff for https://github.com/StrikerRUS/LightGBM/pull/2? Looking at the workflow template, I don't understand how configure.ac is getting changed OR how that change is getting committed.
Notice it was unchanged in the diff for the most recent release: https://github.com/microsoft/LightGBM/pull/3872/files, and recreate-configure.sh does that templating with a temporary configure.ac:
LightGBM/R-package/recreate-configure.sh
Lines 25 to 26 in f879018
Just like it is suggested in https://github.com/microsoft/LightGBM/tree/master/R-package#changing-the-cran-package (Linux or Mac section):
How this process should be done right? |
|
"Edit If the only thing that is changing is the version of LightGBM, you do not need to touch |
OK, got it. What should be touched then? |
you don't need to manually edit any files. Just running |
Ah, seems I'm getting what's going on! Just manually edit |
|
Yep! Just manually edit Sorry if the instructions weren't clear. When I wrote https://github.com/microsoft/LightGBM/tree/master/R-package#changing-the-cran-package , the main audience I had in mind was contributors making changes like #4131, not our usual maintenance like doing a new release. If there are changes to the language there that you think would make it more clear, I'd be happy to review those as part of this PR. |
StrikerRUS
changed the title
StrikerRUS
force-pushed
the
r_configure
branch
2 times, most recently
from
e86cde1
to
bf84c94
Compare
StrikerRUS
changed the title
|
OK, @jameslamb please take one more look. I hope I understood it right this time. |
There was a problem hiding this comment.
This looks great, thanks! The diff in https://github.com/StrikerRUS/LightGBM/pull/1/files#diff-dadfc37f082f218f0e2f0adfc1affc7c55044bcdf8dff110aab9cdfda842d6f3 looks exactly like what I'd expect. Just one minor grammar comment, but I think the workflow is perfect.
I think it is ok that the commit from this bot won't trigger a new CI build. Usually on release PRs (the main place where we update configure) there are additional commits after updating configure that don't have any impact on configure.
For cases like #4131 , this will have to be something that we "just know" to re-run if the contributor doesn't manually regenerate configure. I don't think that's problematic at all, as that file very rarely receives updates that are not part of the release process: https://github.com/microsoft/LightGBM/commits/master/R-package/configure.
| run: | | ||
| git config --global user.name "GitHub Actions Bot" | ||
| git config --global user.email "githubactionsbot@users.noreply.github.com" | ||
| git add "./R-package/configure" |
There was a problem hiding this comment.
I can tell your main OS is Windows because of your defensive use of quotes 😛
|
Thanks!
Great!
It does with the latest code! 🙂 Thanks to
Yeah, unfortunately, GitHub Actions cannot push into forked repos. So we can't do anything with it. |
Co-authored-by: James Lamb <jaylamb20@gmail.com>
|
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. |
Closed #3283.
Refer to #4117 (review) for the initial discussion.
Refer to https://github.com/StrikerRUS/LightGBM/pull/1 for the usage example.
Push from a workflow won't trigger our regular workflows:I don't think that this a problem for us, because anyway we need a follow-up commit with reverting back to~~VERSION~~inconfigure.ac, right?