[Checkers] Declarative specification of qualifier hierarchy

[Mahmood Ali]

On Mar 28, 2008, at 3:17 PM, Matt Papi wrote:

> Is your idea of the TypeRelation significantly different that what  
> is there now? E.g., what would the code for the Javari relation  
> ideally look like under that scheme? It might also help if Mahmood  
> could share the work he's done on the graph relation so far (as I'd  
> suggested in previous message) so you can get a better idea of what  
> the that scheme would look like.
>
I am sorry for not sharing the code.  I was using Internet cafes  
without access to the code.  I will check-in my code for review this  
evening the latest.

>
>>> As a side-note, CustomChecker assumes that the negative annotation  
>>> is a supertype of the positive annotation, which is false  
>>> ("Plaintext/Encrypted", Trusted/Tained, etc).
>
> That's not quite right. The CustomChecker supports two different  
> schemes. [...] Positive and negative are not the best names (they  
> come from the type notation sometimes seen for nullness where T+ =  
> nonnull T and T- = nullable T), though I don't think they're used in  
> any of our documentation.

Thank you for clarifying this.  I was under the impression that we  
wanted to support Plaintext/Encrypted conflict relation, and thought  
that the positive and negative thing was about this.

- Mahmood