attribute data structure rewrite

[Zack Weinberg]

Andrew Pinski <pinskia@physics.uc.edu> writes:

> Shouldn't name be redefined as char* as you don't really need a full 
> IDENTIFIER_NODE?

Or better still, as code numbers (with an identifier->number mapping
through the identifier hash table - as Joseph correctly points out,
the identifiers get allocated during parsing anyway).  Currently we
have to do string comparison to decide which attributes are which
(since any attribute can be written either as "name" or as "__name__").

zw