Closed ns-rse closed 1 month ago
R 4.4.0 is very new, and it seems there's an internal dependency independent of the Workbench that is causing the segfault (this wouldn't be any specific Workbench code that is causing this, rather unmatch builds of lower level libraries outside our control). I suggest waiting until this issue may be resolved in R itself or the library that is conflicting.
I suspect this is related to other segfaults that have been observed elsewhere over the last few days. I have traced those issues to a problem when pkgdown
encounters a few mathematical symbols in source Markdown while converting to HTML. Unfortunately, I have not been able to figure out exactly what is going wrong and why, but I am hopeful for a response to the linked issue to move things forward.
Great find Toby, and thanks for raising the issue with the pkgdown developers!
Thanks @tobyhodges and @froggleston I suspected it was lay much deeper than the Workbench but wasn't sure where to look.
EDIT : I see a fix has been merged into downlit already.
As I understand it, the problem will disappear from the lesson builds once the new version of downlit has been approved and published on CRAN.
Which shouldn't be too long by the looks of it.
Thanks for identifying the root cause and reporting in the correct place @tobyhodges :+1:
On CRAN now!
Confirming that my lesson builds are now succeeding. I think this issue can be closed.
Updated to downlit-0.4.4
and my site that had previously failed builds fine now too.
Thanks @tobyhodges and @hadley :+1:
I've encountered a segmentation fault when attempting to render pages under R-4.4.0.
I've attempted to identify where this occurs using valgrind
...and it looks like its coming from
libR.so
but I'm unsure.This is under Arch Linux and I can't downgrade to
R-4.3.3
because it has a dependency on icu-74 which is now outdated and downgrading would break a bunch of other dependencies.