[bitc-dev] fixint and bool

Swaroop Sridhar swaroop at cs.jhu.edu
Sat Aug 13 18:01:08 EDT 2005


Removing #t and #f from reserved words has a bad effect for fixint.
Now, the parse rule for should be changed to:

fixint: '(' intlit intlit ident ')'

And, error reporting has to be delayed until after symbol resolution. 
Also, if #t and #f are ever rebound, I don't know how fixint should behave.

I am not saying this is situation impossible to handle, but it would be 
nice it fixint were to be purely syntactic. So we should probably 
consider one of the following:

i) (fixint 32 32 0) , that is 0 or 1 for unsigned/signed
ii) Leave bool as a primitive type
iii) Introduce new syntax #\{unsigned} #\{signed}

Swaroop.


More information about the bitc-dev mailing list