RFC: target macro->hook migration idea

[Zack Weinberg]

"Joseph S. Myers" <jsm@polyomino.org.uk> writes:

> You have appropriate performance measures for this?  Needing e.g. always
> to call a function to get the value of a target byte from a string when
> almost all compilers actually have target bytes the same as host bytes
> _seems_ inefficient but the question is: which target macros are used in
> functions that show up significantly on a profile and what is the cost of
> replacing those by structure elements?

I have no hard numbers to prove this, but I have never seen any of the
routines that would be affected show up anywhere near the top of a
profile.

zw