Reference resolving broken
Reference resolving is broken.
The resolve strategy gets passed the actual string node at the cursor position, instead of the surrounding constructor. This seems to be caused by revision 21312.
Submitted by Nathan Bruning on 5 November 2010 at 21:03
Issue Log
I suppose the behavior did change there. What it used to do was just get the parent of a term if it did not have a constructor. So instead of
"foo"
you’d getId("foo")
orCall(..., "foo")
. Now, you only get the parent if it does not have any other children (or if it’s the same lexical construct). So"foo"
becomesId("foo")
but notCall(..., "foo")
.All in all, you could argue that the new behavior is better, or even that the old one is better. But it wasn’t really my intention to change it, so I’ll change it in
r21338
so at least for this scenario it has the old behavior.
Note that the behavior is changed again for Spoofax/358.
Log in to post comments