[ERR5RS] Rationale and ease of use

[Harold Ancell]

At 03:40 PM 9/8/2007, Lynn Winebarger wrote:
>On 9/6/07, David Rush <<mailto:kumoyuki at gmail.com>kumoyuki at gmail.com> wrote:  
>On 9/6/07, Harold Ancell <<mailto:hga at ancell-ent.com>hga at ancell-ent.com> wrote: 
>> At 10:39 PM 9/5/2007, Lynn Winebarger wrote: 
>> >would be willing to make searchable versions 
>> >of the email archives of R6RS (editors), R6RS-discuss, and even the 
>> >old rnrs mailing lists available through scheme-punks. 
>> 
>> I'd certainly be happy to do the work. 
>> 
>> David?
>What does this provide that a google search like
>    "library site: <http://www.r6rs.org>www.r6rs.org"
>doesn't? I'm not opposed, mind you, but I guess I don't understand 
>what is being asked...
>
>
>     I assumed Will had something specific in mind, but I am not sufficiently familiar with mailing list archiving software to know if there is a good solution. 
>    Even a simple search box for the three lists would be handy.  I think a version of the archive with better thread tracking would be ideal

As long as it's fast to read---I really hate the systems that require
one *mouse click* per message, even if I have a local copy.

>(the rrrs archive seemed easily confused, whereas threads in the two r6rs lists get interrupted monthly).  It's not worth moving mountains to achieve, but if it's easy then it's probably worth it. 

I've got a lot of man-hours available now for a task this simple;
I think enough people would like it---and no one would dislike it---
that I will look into Real Soon Now.

People are welcome to send me URLs or the like by private email.

All I need to know is if David can run fairly arbitrary software on
his server, and ... well, for space/time sizing, I need to look at
how much "data" there is, probably not much by current standards.

(Warning, nostalgia trip ahead with a perhaps useful observation at
the end.)

What can you say when L2 caches are frequently 1MB or larger---back
in I think the late '60s or so MIT made a radical proposal to DARPA,
which people said was beyond the state of the art: procure a 1MB (of
9 bit bytes) core memory system, one full PDP-10 address space for
MIT-AI.

When I showed up in '79, it was known as the "Fabra-tek Crufty" (sp.),
and people regretted that the building had windows you couldn't open
to dispose of stuff ^_^.  By then, MIT-MC---an ECL KL-10, the fastest
PDP-10/Decsystem-20 CPU ever fielded, about 2 times a 11/780, procured
with DOE money to run Macsyma---had 6MB, and was nice and shiny compared
to the 3 first generation RDL KA-10s (MIT-AI, ML, and DM, the latter the
homes of Macsyma and Zork respectively).

To tie this back to something useful, the great efforts that went into
getting the best LISP performance out of such tiny by today's standard's
machines (especially to make Macsyma run as fast as possible) means that
LISPs are well positioned when someone throws sand in the face of dynamic
languages.  We can point out our higher performance implementations are
much closer to C/C++ than to Java/Perl/Python/Ruby....

And we need every advantage we can get or legitimately claim.

                                        - Harold