> From: Johan Vromans
> Dave Mitchell <davem@iabyn.com> writes:
>
> > It's these sorts of exceedingly non-obvious gotchas that
> makes me very
> > wary of messing with the debugger.
>
> ... or that make it painfully clear that this piece of
> software requires
> good documentation and/or redesign.
Speaking of redesign.
I think that no-one stops anyone on redesigning the debugger,
given that this newer redeisgned and better debugger will coexist with
good working one... (at least for the transition period)
If the redesigned debugger will be in strict from the very beginning - and
of course "strict" forgiving to user input - then - good. Definitely good.
Speaking of documentation improvement.
Until this redesigned version of debugger appeared, a very useful
documentation addition would be to clearly mark speed-critical loops
of the current debugger (these are 1 or 2 I believe)
I have strong impression that in such critical places even these blocks:
{no strict 'refs'; blablabla;}
are not good, because of extra manipulations with stack.
(I got this impression from ptkdb and I believe it is still true)
Vadim.
Thread Previous
|
Thread Next