Re: experimental new dump

[Thomas Lauer <thomas.lauer@...>]

Luiz Henrique de Figueiredo <lhf@tecgraf.puc-rio.br> wrote:
> > Marginal impact on size.
> > Marginal impact on load time.
> 
> Thanks. This is consistent with my few tests. The new dump should
> produce smaller files only for scripts that have many functions that
> share many strings. Perhaps this is not as common as I thought. So, it
> seems that we're back to the simple solution, which is the current one.

I second that. I have seen reductions in the byte code file sizes but
alas nothing sensational;-) Load time differences were not measurable at
all. (Then again, timers under Windows are crap.)

(A funny aside is that my version of the compiler will optionally
compress the byte code. In ALL cases I tested (mind you, just a
handful), the resulting compressed file was actually BIGGER than with
the old version. Here is a typical case:

byte code old: 53714; same compressed: 26211
byte code new: 51565; same compressed: 26322

Not sure what's going here...)

-- 
cheers  thomasl

web : http://thomaslauer.com/start