Szelethus added inline comments.

================
Comment at: clang/lib/StaticAnalyzer/Checkers/StreamChecker.cpp:857-868
+    // FEof and other states are possible.
+    // The path with FEof is the one that can continue.
+    // For this reason a non-fatal error is generated to continue the analysis
+    // with only FEof state set.
+    ExplodedNode *N = C.generateNonFatalErrorNode(State);
+    if (N) {
+      C.emitReport(std::make_unique<PathSensitiveBugReport>(
----------------
balazske wrote:
> Szelethus wrote:
> > Szelethus wrote:
> > > Ugh, tough one. I think this just isn't really *the* error to highlight 
> > > here, but rather that its bad practice to not check on the stream after 
> > > an operation. But I'm willing to accept I might be wrong here.
> > Actually, I take this back. If we had `NoteTag`s to explain that "this 
> > stream operation failed and left the stream file position indication in an 
> > indeterminate state", this would be great.
> Adding of `NoteTag` and other bug path improvements are planned in the next 
> changes. (The message text could mention that error check was probably 
> forgotten. But the faulty function can be called despite the error check was 
> done.)
Cool! Proceed as its most convenient, no need to rush it while the checker is 
in alpha.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D80018/new/

https://reviews.llvm.org/D80018



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

Reply via email to