[R-package] [docs] fix warnings in pkgdown site building

[jameslamb]

This PR fixes the warnings from the last outstanding item in this comment: #1143 (comment)

Warning message:
Topics missing from index:
* lgb_shared_params
* lightgbm

To confirm, I ran:

Rscript build_r.R
cd lightgbm_r
Rscript -e "pkgdown::build_site(install = FALSE)"

re-using this branch from #3072 since it's already enabled on readthedocs.

[jameslamb]

oh @StrikerRUS was fix/remove-devtools already disabled at readthedocs? Do you mind re-enabling it?

[StrikerRUS]

@jameslamb

oh @StrikerRUS was fix/remove-devtools already disabled at readthedocs? Do you mind re-enabling it?

Yeah, I removed the branch and disabled RTD builds right after merging previous PR. Sorry! Just enabled builds again.

[StrikerRUS]

I can confirm that the warning has gone from build logs and lightgbm() - Train a LightGBM model item has appeared in the R API reference.

Cool, thanks!

[StrikerRUS]

All recent macOS R-package builds at Travis are stuck at

* checking whether package ‘lightgbm’ can be installed ...

No output has been received in the last 10m0s, this potentially indicates a stalled build or something wrong with the build itself.

Hope it is temporary problem...

[jameslamb]

@jameslamb

oh @StrikerRUS was fix/remove-devtools already disabled at readthedocs? Do you mind re-enabling it?

Yeah, I removed the branch and disabled RTD builds right after merging previous PR. Sorry! Just enabled builds again.

no problem! I should have asked you to keep them.

[jameslamb]

All recent macOS R-package builds at Travis are stuck at

* checking whether package ‘lightgbm’ can be installed ...

No output has been received in the last 10m0s, this potentially indicates a stalled build or something wrong with the build itself.

Hope it is temporary problem...

😬 I just tried rebuilding, let's see

[StrikerRUS]

I should have asked you to keep them.

OK!

😬 I just tried rebuilding, let's see

Already tried 3 times with master... 😢

[jameslamb]

😬 Travis status looks fine, no ongoing incidents... https://www.traviscistatus.com/

[jameslamb]

Just added 89c6b3c to see if we can get better logs and identify what is breaking in the Mac R environment on Travis. I'll revert that commit before we merge this, just using it for investigation.

[StrikerRUS]

Just added 89c6b3c to see if we can get better logs and identify what is breaking in the Mac R environment on Travis.

... and it just passed all checks! 😄

I guess building without logs is just near 10mins.
Probably we need some dummy output to prevent Travis from terminating our build:

https://docs.travis-ci.com/user/common-build-problems/#build-times-out-because-no-output-was-received

travis-ci/travis-ci#4190 (comment)

[jameslamb]

WOW! I wonder if this is somehow related to how we switched all print() to message() in install.libs.R?

[jameslamb]

I think it was the print() to message() change in #2965 . Looks like print() prints to stdout and message() to stderr:

[jameslamb]

I documented the issue with timeouts in #3091, and just added a PR for it in #3092

[StrikerRUS]

Great investigation!
Can we get back to print() for the sake of speed?

[jameslamb]

Great investigation!
Can we get back to print() for the sake of speed?

Sorry, I wasn't clear....I don't think our Mac builds are slower now because of message(). I was thinking that since message() prints to stderr, maybe changing print() to message() increased the likelihood that we'd go 10 minutes with no new stdout.

I think that we should:

• keep message() in install.libs.R, agreeing with @Laurae2 that it's the more appropriate choices for package installation messages
• separately investigate why the Mac builds for the r-package have gotten so much slower in the last few days

[jameslamb]

I just rebased to master to get the fix from #3092 . Assuming this builds successfully, I'll merge it and then work on replacing build_r_site.R per #1143 (comment)

[github-actions]

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.