You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think that most segmentation faults that users experience are confined to rsession. Perhaps RStudio could install a signal handler that just gracefully restarts the R session on a segfault, instead of crashing the full IDE?
An almost too-good-to-be-true feature would be if the signal handler did something more, like attach to the rsession with lldb/gdb and print a traceback before restarting, or give an option to drop into the gdb console. This might be as simple as sending a command to terminal like lldb -p <rsession pid>
The text was updated successfully, but these errors were encountered:
Great suggestions, thanks for filing that. We should be offering an option to restart rsession on SEGV. Launching gdb or lldb would be nice to have, or maybe there's a way to have the user direct it to save a core file? The stack is way more useful once we have added back symbols.
In the meantime, to be sure everyone is aware, there is already a signal handler for SEGV that is installed when /etc/rstudio/crash-handler.conf has the enabled flag. Here's how I configure it:
I think that most segmentation faults that users experience are confined to rsession. Perhaps RStudio could install a signal handler that just gracefully restarts the R session on a segfault, instead of crashing the full IDE?
An almost too-good-to-be-true feature would be if the signal handler did something more, like attach to the rsession with lldb/gdb and print a traceback before restarting, or give an option to drop into the gdb console. This might be as simple as sending a command to terminal like
lldb -p <rsession pid>
The text was updated successfully, but these errors were encountered: