NoQ added inline comments.
================
Comment at: clang/lib/StaticAnalyzer/Core/BugReporterVisitors.cpp:364
QualType T = PVD->getType();
- while (const MemRegion *R = S.getAsRegion()) {
- if (RegionOfInterest->isSubRegionOf(R) && !isPointerToConst(T))
- return notModifiedDiagnostics(N, {}, R, ParamName,
- ParamIsReferenceType,
IndirectionLevel);
+ while (const MemRegion *MR = S.getAsRegion()) {
+ if (RegionOfInterest->isSubRegionOf(MR) && !isPointerToConst(T))
----------------
Charusso wrote:
> `R` was cool, that is our unspoken naming convention. What about
> `CallR`/`CallRegion` and possibly `CallSVal`? If I am right usually the
> `BugReport` object is named `BR` because of our regions.
My unspoken (well, not anymore, i guess) naming convention usually goes like
this:
* `B` for `BinaryOperator` (emphasis on binary-ness, `U` for unary so they
don't conflict)
* `R` for `BugReport` (emphasis on being a report)
* `BR` for `BugReporter` (for the lack of better name),
* `MR` for `MemRegion` (specific classes of regions go like `VR`, `FR`, `TR`,
`SR`, or, well, `BR`)
* `BRC` for `BugReporterContext` (emphasis on being a context)
Also `V` for `SVal` (emphasis on value) because `S` for `Stmt`.
================
Comment at: clang/test/Analysis/no-store-suppression.cpp:3
+
+// expected-no-diagnostics
+
----------------
Charusso wrote:
> Could you inject a link for the diff or copy the information for further
> improvements why no diagnostic happen?
Links to diffs are rarely included in the source code because they're usually
two clicks away anyway: just do git blame and see the bottom of the commit
message.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D60107/new/
https://reviews.llvm.org/D60107
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
