NoQ added a comment.

> however as far as I see this will mean the LoopUnroller AST matchers can't be 
> reused unless I change them.

Hmm. What is the problem here? I'm expecting library linkage issues (i.e. 
libAnalysis is not linked to libASTMatchers directly), but i guess it'd better 
to have the whole thing in libAnalysis then (Peter, WDYT?). Or maybe we could 
move the code around a bit, so that we're still in libStaticAnalyzerCore but at 
roughly the same moment of time.

Because you're duplicating the solution for exact same problem, using different 
technology, and this problem is not trivial (eg. you still need to consider 
reference-type declstmts and initializer lists which are not taken into account 
in your solution yet) so i'd dislike the idea of duplicating it.



================
Comment at: lib/Analysis/CallGraph.cpp:104
+
+  void markModifiedVars(Stmt *S) {
+    // Increment/Decrement, taking address of variable
----------------
A hint, even if we don't use visitors: the whole point of having a visitor is 
about not needing to make a pattern-matching (switch or sequence of dyn_casts) 
by statement kind yourself, like you do in this function. You already have 
VisitUnaryOperator, VisitBinaryOperator, VisitCallExpr, etc., so you can add 
the code there.


Repository:
  rL LLVM

https://reviews.llvm.org/D37897



_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to