Charles Oliver Nutter

Note that this is also causing "issues" when exceptions are raised from symlink-pathed files, as seen in #5436. Unclear how critical that particular case is, though, since it was only...

Remaining failures as of today: * Kernel#require_relative with a relative path ($LOADED_FEATURES) with symlinks does not canonicalize the path and stores a path with symlinks * Kernel#require ($LOADED_FEATURES) with symlinks...

We would need to review what changes happen if we update 9.3 resolv.rb to the latest. As long as the changes are minimal and nothing breaks, we could also make...

Thanks again for being on top of this... I will merge this to master, but we should figure out a separate PR for 9.3 so we do not pull in...

To clarify... fully pulling in the latest resolv.rb might be just fine! I just want us to review the changes before we commit to it, since we have had to...

@colinsurprenant Thank you for the extra sleuthing! I grabbed a thread dump from the `read` version, and it is indeed still sitting there happily waiting: ``` java.lang.Thread.State: RUNNABLE at com.kenai.jffi.Foreign.invokeN3O1(Native...

Looking at the CRuby code in `io_bufread`, it does indeed go to `read` before it ever tries to `select`...so this needs more investigation. I am now thinking that we are...

@ccutrer Thanks for the additional example! The only fix I can think of here would still be to use a native pipe for `IO.pipe`, if indeed that is the type...

Interesting. The logic for invoking methods has changed over time, but I'm not sure what part of that would trigger a SecurityException that did not happen before. Perhaps the bundle...

cc @splatch