RFH: ICE on valid code...

[Tom Tromey]

>>>>> "David" == David Daney <ddaney@avtrex.com> writes:

David> Thanks for making me correct my laziness.  You can now find this
David> backtrace as well as a simplified commandline to recreate the problem in
David> the bug report.

Unfortunately, as you surmised in the PR, the stack trace isn't very
useful in this case.  I took a quick look at this.  It is trivial to
"fix" the ICE by papering over it, but given that the ICE occurs in
<clinit> processing, and given that the incorrect error messages also
relate to this, I would say that something is going wrong earlier
while building <clinit>.  More than that would require real
debugging...

Tom