rc and smf

Bill Hacker wbh at conducive.org
Thu Feb 24 13:11:47 PST 2005


Chris Pressey wrote:

On Fri, 25 Feb 2005 03:46:40 +0800
Bill Hacker <wbh at xxxxxxxxxxxxx> wrote:

Erlang is largely concept, and near-as-dammit uniquely Ericsson.


To be fair, it's a pretty nice language. 
Its a *lovely* tool - for what it is good at.  My first impression was 
that it could address a *lot* of challenges - and by no means limited to 
telco.

BUT - perhaps as with the 'Plan 9' concept of aggregating all detectable 
resources one chooses to invite to the party, Erlang's ability  to move 
work around between available resources comes from a model that imposes 
a price that can't entirely be shed when/where not needed.

I've written a construction
cost-estimating program in it, and it's a breeze to maintain.  Very few
of the concurrency features are fully exploited by my application, of
course, but even if they were taken out completely, I would find the
remainder to be one of the most comprehensible functional languages
available.
Agree that - there have been times I wish it could / had already 
replace(d) 'C' <g>

But - speed, universality....

That's not to say it doesn't have its warts, because it has plenty.

But that's neither here nor there...
Exactly.

We 'passengers' are wasting too much developer time on what seems to be 
an inexhaustible well of irrelevant distractions...

Best I go off and look at a simplified (not complexified) startup 
methodology so I don't have to modify or replace /etc/rc all the time. 
Stock version has whiskers, and it's conditionals conditionals have 
. .conditionals....

Plain text, and table-driven, just as fstab is, perhaps with similar 
'pass' numbering, and simple major-key, minor-key grouping to handle pre 
and post preferences per (userland) daemon. The sequence you set is the 
sequence toor uses. Startup and shutdown NOT just the reverse of each 
other..

The added layer of rc.subr has always smacked of Win-bloat to me.

And no 'jihad' if no one else likes it.. <g>

Best,

Bill





More information about the Kernel mailing list