Repository navigation
Fix R runtime - #348
Fix R runtime#348
Conversation
907bd7c to
f5906a8
Compare
3eac847 to
d1564d7
Compare
|
You can see that d1564d7 shows the problematic behavior. Also for some reasons, the conda tests fail, that's why I temporarily deactivated all tests except the R tests (for speed). |
|
I can confirm that this PR runs successfully on all my hooks as I have created an integration test with it in my fork: lorenzwalthert#1 |
|
Still fails on the conda test... |
|
yeah I'm trying to upgrade conda to fix that. for now I've rolled back to the previous image |
|
Thanks for the quick rollback response. |
|
I still see failures when pushing just now. Seems like pre-commit.ci did not pick it up yet? |
|
I haven't had a chance to deploy it yet. had some unrelated issues last night to resolve. should be in a couple hours and then I'll let you know here! |
|
No worries, tomorrow is also fine or any time you get to it. A heads up here would be appreciated. Also, do you have an indication of how many runs this broke? |
|
I've just deployed the newest image |
|
Hooray ,it worked. 🥳 |

#346 removed libblas3 and liblapack3 from the apt list. R doesn’t need those, but liblapack3 depended on libgfortran5, so removing it also removed libgfortran.so.5. R’s stats.so links against that library, so R now fails when calling the {stats} package, a base R package (i.e. shipped with R, like the python standard library):
This breaks most language: r hook environments (renv::install fails while loading stats). We first add a call to {stats} package to demonstrate the behavior in the R tests and later add libgfortran.so back to the main image.