On 10/13/07, Charles Oliver Nutter <[EMAIL PROTECTED]> wrote: > > Keep the compiler in mind; there's no "node" at which to store the > symbol, and I'd rather not have to add more caching code at the compiled > script level if possible. >
The change for 4580 does cache (in StandardASMCompiler#createNewSymbol), but it stores the id rather than the symbol itself. So my point is if you're going to do that, you might as well store the symbol. If you don't want to cache, I'd say revert 4580 and wait for the new concurrent symbol table, which performs about the same in bench_symbol without caching anything. -Bill
