forward references in typedefs

[Bill Rust]

In article <18729 at mimsy.UUCP> chris at mimsy.UUCP (Chris Torek) writes:
>In article <24CB9E07.9547 at marob.masa.com> cowan at marob.masa.com (John Cowan) >writes:
||... As someone (Chris Torek?) said earlier in this group, "all 'struct'
||pointers must 'smell' the same."  If this behavior is not guaranteed,
||forward references to structs would be unimplementable in one pass.
|
|It was Doug Gwyn, but this is otherwise correct.
|
||Ufcawss, if you talk to the people who wrote VMS C, they'll tell you that
||all one-pass implementations of C are unacceptable!  (This has something to
||do with generating good code for the 'switch' statement.)
|
|They are both right and wrong.  One-pass code generation means there
|are few opportunities for optimsiation; but generating good switches
|is easy.  Simply emit a branch at the top of the switch, code at each
|of the labels, a branch at the bottom, and then generate the code
|that actually implements the switch.  E.g., given
|
| [Clever example deleted]


Excuse me. I am not a compiler writer but even I know that there are
two ways to implement switches: jump tables and compare and execute
when equal.  Jump tables (ie jmp switch_list[ix] where ix is some easy
transformation of the switch variable) are much more efficient when
the range of values that the switch variable can assume is limited (ie
switch variable in range of 1 to 20 and it assumes 75% of values). The
only way to tell if a jump table is better than compare and jump is to
see what the range of the switch variable is and how many values it
actually assumes. This is very difficult to do in a one pass compiler.
The previous correspondent completely ignored this in his response.

Bill Rust (wjr at ftp.com)