[C++, diagnostic] PR 34611

[Gabriel Dos Reis]

On Fri, 4 Jan 2008, Paolo Carlini wrote:

| Hi all, hi Gaby,
| 
| for this diagnostic problem, I have this patch, along the lines
| indicated by Jakub: indeed, seems correct to me to get to the TREE_VALUE
| in case of a TREE_LIST, because then dump_expr is able to deal with an
| ERROR_MARK, and, as expected, we are back to the 4.2.x informative
| message (note that the problem with an unhandled TREE_LIST occurs indeed
| in the informative message, not in the error string, therefore, I can't
| prepare a testcase).
| 
| Tested x86_64-linux.
| 
| Would it be ok?

Yes, Jakub's suggestion is right.

-- Gaby