From: "Robert Dodier" <robert.dodier at gmail.com>
Incidentally some time ago I looked at what it would take to
get gcl to compile in-memory, and the simplest solution (barring
misunderstanding on my part, a very real possibility) is to
modify gas (gcc assembler).
> Why don't we remove the call to compile under gcl?
OK by me.
But perhaps not so good for other applications (where the uncompiled
lambda might be more expensive in execution) or on other
implementations (where the compiler is faster).
Instead of implementing in-memory compilation you might try advising
the gcl compile to implement caching. If the second-argument lambda
expression is equal to (one of) the previous (several) arguments then
it could return the previous results. This would be compatible with
the ANS definition of compile.