From: ·············@not.for.spam
Subject: Startup delay
Date: 
Message-ID: <387086d2.539527529@news.earthlink.net>
If you write a simple test program using ACL or equivalent,
and generate an executable binary program, which does
very little, such that it should take a millisecond or less, can you
run it from a script hundreds of times per second?  Or is there
a startup delay such that it would take tens or hundreds of
milliseconds for the program to start, so it could only be run
a few times per second?

From: Marco Antoniotti
Subject: Re: Startup delay
Date: 
Message-ID: <lwu2kvpbvx.fsf@parades.rm.cnr.it>
·············@not.for.spam writes:

> If you write a simple test program using ACL or equivalent,
> and generate an executable binary program, which does
> very little, such that it should take a millisecond or less, can you
> run it from a script hundreds of times per second?  Or is there
> a startup delay such that it would take tens or hundreds of
> milliseconds for the program to start, so it could only be run
> a few times per second?

(Common) Lisp is the ultimate scripting language.  How many times do
you start Emacs in a week? :)

Cheers

-- 
Marco Antoniotti ===========================================
PARADES, Via San Pantaleo 66, I-00186 Rome, ITALY
tel. +39 - 06 68 10 03 17, fax. +39 - 06 68 80 79 26
http://www.parades.rm.cnr.it/~marcoxa
From: Tim Bradshaw
Subject: Re: Startup delay
Date: 
Message-ID: <ey3r9fzcj9j.fsf@tfeb.org>
* not for email wrote:
> If you write a simple test program using ACL or equivalent,
> and generate an executable binary program, which does
> very little, such that it should take a millisecond or less, can you
> run it from a script hundreds of times per second?  Or is there
> a startup delay such that it would take tens or hundreds of
> milliseconds for the program to start, so it could only be run
> a few times per second?

I recently posted some figures for a couple of Lisps (not ACL) to this
newsgroup.

On a 333MHz ultrasparc with enough memory, gcl 2.3 is around 0.03 secs
to run a null program

clisp (recent version) seems to be about 0.08

cmucl seems to be about 0.2

gcl startup is comparable to perl, which seems to be quite widely used
for scripting.

Allegro seems to be substantially slower than these, but I've not
played with arguments &c enough to be really sure (I used cl -Q -qq
-kill).

I think for most purposes I'd tend to have a persistent Lisp which I
talked to down a network connection, but that's really because I'm
interested in applications which have a lot of complex persistent
state which I want to keep in memory all the time -- gcl and clisp
look more than adequate for small perl-style scripts in terms of
startup, and others may be too.

--tim
From: ArolAmbler
Subject: Re: Startup delay
Date: 
Message-ID: <20000109075014.12489.00000591@nso-fa.aol.com>
Why not start the program, with the input read from a pipe.   When the data
arrives, there should be very little overhead getting it there and processed.