And how much bandwidth has been wasted by these verbose instructions
on how to write emails to this list?
I considered this, too; but I figured an email could convince
a few other people to quote less, it might be an overall win...
It seems that regularly threads turn to a discussion about how to
use the mailing list, and I wonder just how useful that is.
I think it's a useful reminder because this translates to
knowledge about performance that can help Ruby (or any
computing).
As a sometimes ruby-core dev; I get complaints about Ruby (and
other things :P) being slow and using too much memory; but much
of that blame is programmers become careless and complacent from
using powerful hardware that they forget users who cannot
upgrade all the time.
I know I wrote some of my slowest code right after I got more
RAM and faster CPUs :>
My hope is that if more people are aware of small everyday
things which affect performance; they can help improve Ruby
and the computing world.
Now, I still regularly use old, slower hardware. I notice
performance problems way more quickly that way.
<git old-timer tangent>
One of the most enlightening reads in the earlier days of
git was this discussion on tabs vs spaces in indentation
and how it affected performance of "git grep":
http://mid.gmane.org/20071018024553.GA5186@coredump.intra.peff.net
(tl;dr: tabs were 16% faster because they're more compact)
Size matters, locality matters. git's design awareness around
those things made git fast to start with; then it also benefitted
from continuous efforts to optimize small bits and pieces here
and there; adding up to huge wins.
</git old-timer tangent>
In contrast, Ruby wasn't designed for speed from the start;
and the internal data structures had poor locality.
However, mainline Ruby made improvements over the years:
* bytecode VM + inline caches improves memory locality
* generational GC reduces scanning of the heap
* more compact and better locality of internal tables
* reducing allocations and exceptions in common code
Sadly, much of the performance improvements made to Ruby gets
negated since Ruby applications get bigger and push more data
every year, too :<
I'm plenty aware git still has performance problems, too;
and performance problems are everywhere you look (including
the lack of compression of this message over SMTP)
···
Joe Gain <joe.gain@gmail.com> wrote: