On 12/9/19 7:06 AM, Paolo Carlini wrote:
Hi,
On 08/12/19 18:51, Jason Merrill wrote:
Hmm, is the change to cp_expr really necessary vs. using
protected_set_expr_location?
Yes, using protected_set_expr_location works fine in this case, I
suppose because we are dealing with expressions anyway plus the cp_expr
constructor from a tree copies the location too. In the below I also
added the thin build_functional_case wrapper, this way consistently all
the build_*_cast functions called by the parser do not use set_location
afterwards. Note, at some point we should also do something about the
build_x_* functions which have been doing that for a while...
Anyway, the below passed testing.
Thanks, Paolo.
////////////////////////////////
OK.
Jason