[OT] Nostalgia
81 views
Skip to first unread message
Hendrik Boom
Sep 6, 2019, 9:41:23 AM9/6/19
to Racket Users
I just noticed DrRacket telling me
Language: racket, with debugging ; memory limit: 128 MB
immediatey reminding me of the 128 K that was available on the first
Lisp system I got any actual use from -- one I implemented myself on an
IBM 360/65. Back then there were rumours that some IBM customer
somewhere in the world had a machine with maximum storage -- 16 MB.
And now we *restrict* ourselves to more RAM than one of the largest
machines then had available.
And I got a lot done on that 128K machine, three orders of magnitude
ess than we hand out to students now.
-- hendrik
Language: racket, with debugging ; memory limit: 128 MB
immediatey reminding me of the 128 K that was available on the first
Lisp system I got any actual use from -- one I implemented myself on an
IBM 360/65. Back then there were rumours that some IBM customer
somewhere in the world had a machine with maximum storage -- 16 MB.
And now we *restrict* ourselves to more RAM than one of the largest
machines then had available.
And I got a lot done on that 128K machine, three orders of magnitude
ess than we hand out to students now.
-- hendrik
Stephen De Gabrielle
Sep 6, 2019, 9:50:00 AM9/6/19
to Racket Users
LOL
128k ! I remember dreaming I had 128K when I was on a 16 K machine.
Recently had to change my limit to 1024 MB...
S.
--
You received this message because you are subscribed to the Google Groups "Racket Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to racket-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/racket-users/20190906134118.ehqjvbvebkgoaqtd%40topoi.pooq.com.
----
Hendrik Boom
Sep 6, 2019, 10:05:58 AM9/6/19
to Racket Users
On Fri, Sep 06, 2019 at 02:49:45PM +0100, Stephen De Gabrielle wrote:
> LOL
> 128k ! I remember dreaming I had 128K when I was on a 16 K machine.
Well, that 128K was a big omprovement on the machine I *first* tried
> LOL
> 128k ! I remember dreaming I had 128K when I was on a 16 K machine.
implementing Lisp on -- an IBM 1620 with 40,000 decimal digits of
memory. Addresss were five digits, so a typical cons cell cost ten
digits, leaving space for 4000 cons cells, less the space the
interpreter took. Proof of concept, maybe, but not very useful.
Lking back, could I perhaps have improved things by using four-digit
addresses? Maybe the extra addressing code would have eclipsed the
savings.
-- hendrik
Neil Van Dyke
Sep 6, 2019, 11:36:03 AM9/6/19
to Racket Users
Last week on HN, a non-student was complaining about having to increase
DrRacket's memory limit 3 times while they were playing with it, so I
pointed out that DrRacket was designed for new students, and suggested
that maybe that memory limit was a good thing for new students.
One of frequent complaints, from generation to generation, seems to be
"kids these days got it too easy". Which, in programming, is not
necessarily grumpy, but concern that a lot of learning opportunity that
comes from working with resource constraints is being missed.
Personally, I think there's a balance, and there's also learning
opportunity missed when you don't have lots of resources. Ideally, a
person gets both kinds of experiences.
As a C and C++ programmer who was then an early Java promoter, I was a
bit concerned about that, at the time. I figured we'd probably move to
Java, and all the students already had use of powerful multiprocessor
workstations. That was one of the attractions of then playing with
programming the Pilot PDA ("https://www.neilvandyke.org/t-map/"), and I
promoted Pilot programming to other students specifically for the reason
of learning to develop with tight resource constraints.
Other Racket relevance: the approach to fitting the toy "route planner"
into a few KB was to use two little DSLs, with a Lisp as code generator
to get around the limitations of macro preprocessing in C. Between
that, the crazy DSLs I made as sets of cpp macros for my compiler (C++)
and robot (C) assignments, and an awful concurrent hierarchical state
machines language that took way too much effort to implement in Java, I
suppose it's not a surprise I later decided to move to Scheme/Lisp for
my main research tools.
Also, copious computing resources becoming available to lots of people
became a concern to some researchers, who no longer felt as privileged
as before. At the start of the dotcom boom, one of my grad school
advisors was already spending most of their time on startups (and there
was some truth to the joke about advisor only wanting MS+IPO students).
They called an off-site retreat for our group, where a big part of the
case was that the university research lab no longer had special
advantages like supercomputers that other people didn't have. (And I'd
previously been horrified to see the storage array cabinet from a
Connection Machine being used as a barkeeper counter, for the lab's posh
sponsor events.).
Today, it's true: I have my own deep neural networks supercomputer in my
living room, for a few hundred dollars, and it's just an ordinary
consumer GPU like children have in their gaming PCs/consoles. Which
makes for "lots of resources" learning opportunities, complementing the
"not enough resources (until you figure it out)" learning opportunities.
DrRacket's memory limit 3 times while they were playing with it, so I
pointed out that DrRacket was designed for new students, and suggested
that maybe that memory limit was a good thing for new students.
One of frequent complaints, from generation to generation, seems to be
"kids these days got it too easy". Which, in programming, is not
necessarily grumpy, but concern that a lot of learning opportunity that
comes from working with resource constraints is being missed.
Personally, I think there's a balance, and there's also learning
opportunity missed when you don't have lots of resources. Ideally, a
person gets both kinds of experiences.
As a C and C++ programmer who was then an early Java promoter, I was a
bit concerned about that, at the time. I figured we'd probably move to
Java, and all the students already had use of powerful multiprocessor
workstations. That was one of the attractions of then playing with
programming the Pilot PDA ("https://www.neilvandyke.org/t-map/"), and I
promoted Pilot programming to other students specifically for the reason
of learning to develop with tight resource constraints.
Other Racket relevance: the approach to fitting the toy "route planner"
into a few KB was to use two little DSLs, with a Lisp as code generator
to get around the limitations of macro preprocessing in C. Between
that, the crazy DSLs I made as sets of cpp macros for my compiler (C++)
and robot (C) assignments, and an awful concurrent hierarchical state
machines language that took way too much effort to implement in Java, I
suppose it's not a surprise I later decided to move to Scheme/Lisp for
my main research tools.
Also, copious computing resources becoming available to lots of people
became a concern to some researchers, who no longer felt as privileged
as before. At the start of the dotcom boom, one of my grad school
advisors was already spending most of their time on startups (and there
was some truth to the joke about advisor only wanting MS+IPO students).
They called an off-site retreat for our group, where a big part of the
case was that the university research lab no longer had special
advantages like supercomputers that other people didn't have. (And I'd
previously been horrified to see the storage array cabinet from a
Connection Machine being used as a barkeeper counter, for the lab's posh
sponsor events.).
Today, it's true: I have my own deep neural networks supercomputer in my
living room, for a few hundred dollars, and it's just an ordinary
consumer GPU like children have in their gaming PCs/consoles. Which
makes for "lots of resources" learning opportunities, complementing the
"not enough resources (until you figure it out)" learning opportunities.
Annaia Berry
Sep 11, 2019, 3:03:32 AM9/11/19
to Neil Van Dyke, Racket Users
Amusingly enough, I've spent a little time playing with a Lisp someone ported to the 128K Color Computer 3.
It is very hard to do much of anything once the interpreter and library is loaded in. To be honest I'm still impressed it runs at all. :D It's a pretty impressive little dialect for an 8-bit machine, even has macros and lambdas, which most attempts lacked entirely.
--
You received this message because you are subscribed to the Google Groups "Racket Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to racket-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/racket-users/8458611a-0bd8-8af3-dab9-dba61b8c2c42%40neilvandyke.org.
Stephen De Gabrielle
Sep 11, 2019, 9:19:34 AM9/11/19
to Annaia Berry, Neil Van Dyke, Racket Users
I remember being interested in Acornsof Lisp for the BBC Micro, but sadly it was too expensive to be seriously considered(£80 iirc). I did manage to get a hold of 'A little smalltalk' and the associated book later, but ended up going down the Racket rabbit hole instead...
S.
To view this discussion on the web visit https://groups.google.com/d/msgid/racket-users/CA%2BHVF0BZPDexfVvkZNuxwKx2qpV4oDkyVBEumQBn70g2qQ_RHQ%40mail.gmail.com.
Annaia Berry
Sep 13, 2019, 2:38:52 AM9/13/19
to Stephen De Gabrielle, Neil Van Dyke, Racket Users
Yeah, one of the inspirations for Heresy is that a while back I went on a big kick for hunting down early Lisps for 8bit and 16-bit micros.
Really most of the 8-bit ones are pretty terrible; extremely limited, missing key features of the language (a lot of them leave out lambdas entirely). The hardware just isn't up to the task, and most use so much memory there's little left for your programs.
There's a few OK early Schemes though for early 16-bit machines though, Gambit crops up on the Amiga, I think there was a pretty good Lisp for DOS that gets a lot of praise, the Mac got the legendary MCL. And if you go all the way back to MzScheme there was even a version of that for later 68k Macs. ;)
Reply all
Reply to author
Forward
0 new messages