Re: Making Guile slower

[Mikael Djurfeldt <mdj at mdj dot nada dot kth dot se>]

Greg Harvey <Greg.Harvey@thezone.net> writes:

> Mikael Djurfeldt <mdj@mdj.nada.kth.se> writes:
> 
> > And, if it turns out that this isn't sufficient, and that people
> > anyway get GC problems with roots here, then---sure---we could perhaps
> > sacrifice 5% of performance in order to achieve greater robustness.
> 
> Waiting for someone to have a bizarre problem that they can't explain
> is *not* the answer (how long have these bugs been hidden in
> guile?). What you'll observe (if you have the stamina to follow it all
> the way through; this'll take a while) is one of your objects suddenly
> becoming another object. So you'll spend a couple of hours looking for
> assignment mistakes through the code, you'll try to figure out why you
> chose to use that cell to create a new object, and you'll end up very
> frustrated; once you discover that it's guile's fault, you'll be quite
> likely to turn to something other than guile (if it happened to me,
> and I didn't know why, I'd certainly be pissed off).

I should have formulated it differently.  What I meant was that I
think robustness is worth a small sacrifice of performance.

(And I have indeed spent many hours of gdb-time inside gc.c trying to
 figure out what happened to my poor heap.  :)

(To be fair against Guile, I should add that the last two years it
 has always been the fault of my application code.)