Conversation
|
|
||
| # find the version of R in play | ||
| find_package(LibR REQUIRED) | ||
| execute_process( |
There was a problem hiding this comment.
Can we check for / report an error here if one occurs? (Basically signal for future-us that --slave is really gone now)
There was a problem hiding this comment.
This will already happen elsewhere:
rstudio/src/cpp/CMakeLists.txt
Lines 469 to 478 in b22f63e
If you're wondering why we don't piggyback on this check, it's because the r/ folder is a separate CMake project which has no visibility of the parent project's variables.
|
Are you going to backport this to the v1.2 branch? |
|
Please make a release because the currently released 1.2.5042 isn't compatible with R-4.0.0. |
|
We're working on it (should be within a few weeks). |
|
I am receiving the same error with 1.3.959 so is this still open? WSL 2, Ubuntu 20.04LTS, 64bit |
|
This should be fixed in the v1.3 patch release: rstudio/src/cpp/r/session/REmbeddedPosix.cpp Lines 109 to 113 in 21b2886 |
This is the minimal change required to get RStudio building against R 4.0+. In R 4.0, the field
R_Slavewas removed fromstructRstartand replaced withR_NoEcho.wch/r-source@f1ff49e#diff-178cb445c6da1cdb9d55c898b6118256R70
Since RStudio references this field, and we would like to be able to build against both R 4.0 and R 3.x (largely for ease of development and testing), we need to refer to the field by the appropriate name.
The irony of using
--slaveto determine whether we need to useR_SlaveorR_NoEchois not lost. R 4.0 (currently) only soft-deprecates--slave, whereasR_Slavehas been simply removed.Related to #5923. We will need to make another (significantly more invasive) change when R fully removes
--slave.