[775] in Coldmud discussion meeting

root meeting help first first in chain previous in chain previous next next in chain last in chain last

Re: fork() vs send()

daemon@ATHENA.MIT.EDU (Sun Aug 20 17:25:03 1995 )

Date: Sun, 20 Aug 1995 15:21:11 -0600
From: 869683 Gillespie Brandon James <brandon@avon.declab.usu.edu>
To: coldstuff@MIT.EDU

< 869683 Gillespie Brandon James drew these hieroglyphs:
< ^^^^^^
< Why the number before your name?

Local accounting uses it, although I'm exhempt from it, so its all a waste :)

< Hmm.. I'm not sure I see the need for it.  The function wouldn't do
< anything new, and a general-purpose word like "send" just doesn't hint
< at looping either.

It isn't as much a need as an optimization.  It is faster for the driver to
iterate a list in the above manner, than for you to with pcode.

I dunno though, overall the win may not be worth it.

< There is a difference.  In a delayed fork, the task doesn't exist
< until it runs, and in a fork-then-pause, it does.  This does make a
< difference.

How do you know this?  I havn't even written it yet.

Besides, whether the task exists or not is irrelevant, it still has to
duplicate the current execution frame.  The task itself is just a pointer
to that frame.  If I were to implement it with delay, it would exist from
the second it was forked in the task list, because it would exists as an
execution frame.