RFC: identifier GC

[Tom Tromey]

>>>>> "Mark" == Mark Mitchell <mark@codesourcery.com> writes:

Mark> Right; I think this is key.  The ugly nocopy bit in the C++
Mark> mangling code was -- at least at one point -- a relatively
Mark> significant win.  I think that if you look at the history,
Mark> you'll find this was something we did in the process of speeding
Mark> up the C++ front end.

FWIW I did go read the original message:

    http://gcc.gnu.org/ml/gcc-patches/2004-08/msg02027.html

My take-away was that the substitution buffer change (whatever that
is) was the known culprit, and that the effect of the other change was
not measured.

Still, I'll get to this.

Tom