Closed Ilia-Kosenkov closed 1 year ago
@yutannihilation , any idea why this is failing? Cannot reproduce locally. I suspect something wrong with the source code on Windows.
I am currently debugging CI, so the latest runs fail on purpose.
Is the long-path support a Windows feature? Maybe we should have this check:
- name: Check LongPathsEnabled
run: |
(Get-ItemProperty "HKLM:System\CurrentControlSet\Control\FileSystem").LongPathsEnabled
there is this link as well https://www.microfocus.com/documentation/filr/filr-4/filr-desktop/t47bx2ogpfz7.html I'm not finding something for making it possible in the runner.
For some reason, it does not fail on R-devel on Windows, which may suggest that R-devel does things differently.
Yes and no. I am fairly confident in the approach I've chosen, but these tests guard us against such path/libraryname problems. And I guarantee you there will be another person using rust_source()
in a weird scenario, so better be safe than sorry :)
R-devel does things differently.
It seems the implementation around file path on Winodws is getting overhauled in R-devel. But, I don't understand the details and how it relates to the difference we observed here... Anyway, good that it passed the CI.
Fixes #234
rust_source()
correctly consumes source code from itsfile
argumentcode
argument)Fix regression introduced in #229
subset_lines_to_fit_in_limit()
, which affectedui_throw()
Fix redirected link in a vignette
Remove temporary value assigned to
warning.length
which was needed to bypass{rlang}
issueAdd rust toolchain name to the display name of the pipeline