Fix for enrich calls bug #141
Conversation
|
Documentation for this branch can be viewed at https://sites.ecmwf.int/docs/loki/141/index.html |
Codecov Report
@@ Coverage Diff @@
## main #141 +/- ##
==========================================
+ Coverage 92.09% 92.11% +0.02%
==========================================
Files 89 89
Lines 16512 16516 +4
==========================================
+ Hits 15207 15214 +7
+ Misses 1305 1302 -3
Flags with carried forward coverage won't be shown. Click here to find out more.
|
awnawab
force-pushed
the
naan-enrich-calls-fix
branch
from
0bc6824
to
5d3461e
Compare
There was a problem hiding this comment.
Many thanks, well-tested fix for a subtle bug.
As discussed offline, the proper way of fixing this in the future would be to improve the representation of interfaces. While the interface node needs to define a scope, it shouldn't really have a parent because it can't inherit symbols.
However, at the same time we likely want to retain the behaviour of registering the symbols declared by the interface in the parent scope's symbol table.
This is a more invasive change, though, and for a future PR.
| for node in [node for intf in self.interfaces for node in intf.body]: | ||
| if getattr(node, 'name', None) == call.name: | ||
| if node.parent == self: | ||
| node.parent = None |
There was a problem hiding this comment.
This breaks with the nested scoping behaviour that we're using normally. However, the root problem is that interface blocks shouldn't really have a parent scope in the first place because they can't inherit symbols, so I'm ok with this temporary solution.
There was a problem hiding this comment.
Filed a follow-up ticket in #147
This PR fixes the bug highlighted in #134 (in the presence of an explicit interface,
call.namestill points to symbol defined by interface even after enriching calls).