Correct path generation #20
Conversation
This reverts the generation of the 'new_path' of the DLL to the version where `normalizePath` calls `file.path`. These paths can then correctly be matched to the normalized paths of the active DLLs also on Windows, so that unloading of active DLL works as intended. The argument 'mustWork = FALSE' to 'normalizePath' gets rid of the warnings which we do not care for as we know that the new path will often not point to an existing file at this point.
|
I'd seen that argument but didn't "click" it would suppress the warning and 'do the right thing'. Nice catch. rhub r-relrease https://win-builder.r-project.org/5K5xSj5W4JN3 |
|
And we're now on CRAN as 0.3.17: https://cran.r-project.org/package=inline So full steam ahead for your next |
I need to stabilise my interface to mixed effects models that I am currently working on and get my test coverage back up - currently it is only 61% as I introduced a lot of new code recently. And testing graphics code is still a pain even with Anyways, thanks a bunch for being responsive, it was a pleasure contributing! |
|
Aside: Visual diffs are a pain. We usual something coarser (MongoTheCat/visualTest -- by Gabor but older) to test with our students but it isn't ideal either. |
This reverts the generation of the 'new_path' of the DLL to the version
where
normalizePathcallsfile.path. These paths can then correctlybe matched to the normalized paths of the active DLLs also on Windows,
so that unloading of active DLL works as intended. The argument
'mustWork = FALSE' to 'normalizePath' gets rid of the warnings which we
do not care for as we know that the new path will often not point to an
existing file at this point.