[cfe-commits] patch: simplify and fix LookupVisibleDecls

Argyrios Kyrtzidis kyrtzidis at apple.com
Tue Mar 27 13:52:28 PDT 2012

On Mar 27, 2012, at 12:21 PM, Nick Lewycky wrote:

> Ping!
> Argyrios, you made the change in this value from 35 to 37. Could you please take a look?

I'll take a look but I didn't just change the value from 35 to 37, this was the full change:

-// CHECK-CC1: ObjCPropertyDecl:{ResultType double}{TypedText prop4} (35)
+// CHECK: ObjCIvarDecl:{ResultType double}{TypedText prop4} (37)

notice that, after the change, autocomplete shows the ivar instead of the property.

> Nick
> On 23 March 2012 22:12, Nick Lewycky <nlewycky at google.com> wrote:
> LookupVisibleDecls does redundant work which was caught by my new assertion (see PR12339) but upon further investigation we found that it's also buggy in some cases. Two new testcases are included.
> The attached patch works by adding a new iterator on DeclContext which exposes all decls that would be visible through DC->lookup() as an iterator, then LookupVisibleDecls just iterates through those. The fix also allows us to delete some code in code completion.
> There is one form of fallout which I don't understand, and I need review for this. In one Objective-C test, the autocomplete priority changed from 37 to 35. It's not clear to me whether that's a bug fixed or a bug introduced, and I'd appreciate some help understanding why prop4 should have been getting the 2-point penalty in test/Index/complete-synthesized.m (and not, say, prop2 which actually is in the base class). It looks from the commit log like that number might not actually be important.
> Please review! This patch has Richard Smith approval, subject to the change to the objective-c tests being approved by somebody else.
> Nick
> <pr12339-2.patch>_______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20120327/831d0d92/attachment.html>

More information about the cfe-commits mailing list