steakhal wrote:

> > I could repurpose conjured symbols for this, and shove this metadata into 
> > it. WDYT about this? @NagyDonat @Xazax-hun
> > I could explore this direction.
> 
> I think that is a promising idea. Perhaps instead of introducing this 
> `InvalidationCause` hierarchy, you could introduce a `ConjurationCause` 
> hierarchy which includes the `InvalidationCause` hierarchy as one branch? 
> (I'm not sure whether this is actually useful, but perhaps worth a thought. 
> Also, you can postpone the development of the non-`InvalidationCause` side of 
> the hierarchy if there are no plans to use it right now.)

I have no idea what the "other" branch would be of that hierarchy. If this is 
about describing how the value was changed (a direct binding was changed to 
another direct binding due to a direct escape) or what caused this change 
(calling an opaque function having this variable potentially mutated, which had 
a second degree impact through some reachable pointer such that we 
pessimistically escape that region) are orthogonal concepts.
I could store these in separate fields of course.


https://github.com/llvm/llvm-project/pull/207155
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to