Re: Storing Bytecode Offset

[Andrea Corallo]

Rocky Bernstein <rocky@gnu.org> writes:

> Andrea also suggested giving up on the most recent offset and keeping
> offsets in the backtrace.

To be precise this is somenthing I mentioned as a possibility if none of
the other perf impact mitigations works out sufficiently.

> We think that this is bad in that the most-recent position is the one
> that is most desired.


Anyway IMO would be probably more fruitful to focus on the byte compiler
preserving location topic as first.  This is has a value by its own
allowing for emitting correct diagnostic *and* is a pre for having the
bytecode offset thing useful.

  Andrea

-- 
akrl@sdf.org