Am 19.02.2011 00:02, schrieb David Murn:
This is modern-day thinking. Modern solutions are to simply throw more
money and hardware at a problem, where older techniques called for using
the same hardware but making the code faster. That then means if you
improve the hardware you get a double-increase in performance. The
problem these days, is that code is moving to less optimized forms,
because the newer hardware can handle it, where if it was written for
old hardware, it wouldnt be getting as bogged down now.
I remember the very same discussions about C vs. assembler more than 20
years ago. You may think about an XAPI implementation in assembler -
probably even faster than C ;-)
Security is all well and good, but if the service is too slow to work
for anyone, what good is an ultra-secure codebase?
Sorry, this kind of discussion is just pointless.
What good is a C codebase that crashes often, invites everyone to
install a rootkit and leaks memory as hell?
Doesnt it make sense then that if people are complaining about the
performance, it might be worth looking at changing? Its not like every
user has to run the java code on their own system, it has to be no more
portable than the postgres server running alongside it. While there may
be cases for using java in a web-app, the same way there are cases for
using flash, I dont think anyone would advocate re-writing xapi in
flash, even if that was the only language they knew.
If someone would provide an XAPI implementation in flash today that just
works, it might even have a chance to replace the current implementation :-)
Regards, ULFL
_______________________________________________
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk
