Geoffrey White

@dbartol are you able to reply to the last point in the thread? One way or another, I am going to be done with this issue next week.

I've just switched this to @dbartol's suggestion: ``` use = def and result instanceof MustExactlyOverlap ``` - performance is still very much improved (though a newer database doesn't match the...

> > we also lose overlaps between things like *p and ?*p in quite a lot of places > > @geoffw0 can you point me to an example of this?...

@dbartol how do you feel about the `*p` and `?*p` overlap changes revealed by the test? (at this point I'd be happy to just undo the changes if we're not...

Merged in recent `main`, fixed formatting, started new DCA run. Update: DCA run successful, no alert changes, analysis time about the same overall.

@dbartol any thoughts on my answer to your question above?

I've added your commit. I'm happy with it. ✨ `cpp/path-injection` on `AcademySoftwareFoundation/openexr` now takes 224s :+1: @jketema perhaps we could have an opinion from someone currently on the CPP team,...

I don't really understand the motivation for `@`. If we're not propagating indirect data properly through models, isn't that an issue with data-flow itself (or the indirections feature) - not...

Thanks, that makes sense. I'll find time to review the code changes at a later point, please ping me if anything is waiting on this.

> I wanted to delay the change note until we were sure this was the format we wanted. That sounds sensible. There's also the information at the top of `ExternalFlow.qll`...