[e-lang] Non-transparent memoization: is this safe?

[Mark S. Miller]

Kevin Reid wrote:
> The inability to keep a limited-size cache is one possible reason one  
> might not want "intern"'s behavior.

The issue isn't whether we want interning behavior. As you've pointed out, 
this is the right way to account for the semantics of importing DeepFrozen 
emakers. These need to intern, in order to export shared identity, in just the 
way Java's ClassLoaders must intern imported classes. The only issues I see 
are 1) whether such interning should be provided by memoize or a separate 
abstraction, and 2) whether we should use interning for memoizing-like 
performance purposes. The answers to these are obviously related. IIUC, 
E-on-CL does both, which I think is a fine answer.

-- 
Text by me above is hereby placed in the public domain

     Cheers,
     --MarkM