Szelethus marked an inline comment as done. Szelethus added a comment. In D75698#1908335 <https://reviews.llvm.org/D75698#1908335>, @NoQ wrote:
> In my head this patch should ideally be reduced to a single if-statement: > "This value is a `SymbolDerived` //therefore// it was produced by > invalidation". Getting to this point already forced me to learn a lot about the inner workings of the analyzer, and I suspect I have quite a bit of work ahead of me still. Infact, I'm not even sure what `SymbolDerived` is. I suspect this approach should be abandoned then? > It's harder than that, of course, because some derived symbols are legitimate > (i.e., values returned through out-parameters). But this is a separate > problem: instead of producing anonymous conjured symbols, invalidation should > produce richer symbols that explain what kind of invalidation has happened > and which specific effect of this invalidation is represented by this symbol. > Or we could instead make better symbols for representing out-parameters. Alright, I've got some learning to then :) Thanks! Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D75698/new/ https://reviews.llvm.org/D75698 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
