30 Oct
2015
30 Oct
'15
11:16 a.m.
From: Fred Cisin
2 or 3 decades ago, the folk controlling lower
division undergraduate
"Computer Science" at UC .. declared, "Assembly language is dead!
Nobody will ever program in it again.", and shifted their program to
Scheme/Lisp.
..
assembly language may no longer be a great career path, but there will
always be need for some levels of hand optimization.
It's worth teaching a bit of machine/assembler language, so that students
understand how computers _actually work_, underneath.
There's a semi-famous incident from a couple of decades back at MIT: they
were teaching a programming course in CLU (an "object-based" language which
contributed many ideas to object-oriented programming). So one assignment was
to write an assembler - which required being able to print octal numbers.
So quite a few of the students wrote 'octal clusters' ('cluster' is CLU
jargon for the collection of routines which know-how/are-allowed to operate
on members of a class), which used normal decimal read and write to do
input/output - and had 'octal add' etc routines which took apart two 'octal
numbers' abcdef, stored as the decimal number abcdef, into their constituent
digits, added them together individually, did the carries, and then put it
all back together. (I am not making this up. This really happened.)
Apparently nobody had ever told them that a number in memory is... just a
number. At which point it became clear that they needed to know a little
more about how a computer actually worked.
Noel
30 Oct
30 Oct
11:34 a.m.
On 2015-10-30 2:16 PM, Noel Chiappa wrote:
Pretty sure this same thing is repeated somewhere with every generation
of programmers :-)
But I'm not sure why anyone is shocked at "60 year old languages" being
required. People are constantly telling us to use a FORTY year old
language for crypto and systems work as if nothing had been learned since.
--Toby
From: Fred
Cisin
2 or 3 decades ago, the folk controlling lower
division undergraduate
"Computer Science" at UC .. declared, "Assembly language is dead!
Nobody will ever program in it again.", and shifted their program to
Scheme/Lisp.
..
assembly language may no longer be a great career path, but there will
always be need for some levels of hand optimization.
It's worth teaching a bit of machine/assembler language, so that students
understand how computers _actually work_, underneath.
There's a semi-famous incident from a couple of decades back at MIT: they
were teaching a programming course in CLU (an "object-based" language which
contributed many ideas to object-oriented programming). So one assignment was
to write an assembler - which required being able to print octal numbers.
So quite a few of the students wrote 'octal clusters' ('cluster' is CLU
jargon for the collection of routines which know-how/are-allowed to operate
on members of a class), which used normal decimal read and write to do
input/output - and had 'octal add' etc routines which took apart two 'octal
numbers' abcdef, stored as the decimal number abcdef, into their constituent
digits, added them together individually, did the carries, and then put it
all back together. (I am not making this up. This really happened.)
Apparently nobody had ever told them that a number in memory is... just a
number. At which point it became clear that they needed to know a little
more about how a computer actually worked.
Noel
11:35 a.m.
On Oct 30, 2015, at 2:16 PM, Noel Chiappa <jnc at
mercury.lcs.mit.edu> wrote:
...
Apparently nobody had ever told them that a number in memory is... just a
number. At which point it became clear that they needed to know a little
more about how a computer actually worked.
That reminds me of an exchange I had with a younger colleague who was writing a document
describing how to do incremental TCP checksum update correctly. He was tripping over the
issue of negative zero. I told him the answer simply was to subtract the complement with
end around borrow. He didn't believe me and ended up proving it by exhaustively
testing every case.
Sigh.
paul
12:15 p.m.
On Oct 30, 2015, at 11:35 AM, Paul Koning
<paulkoning at comcast.net> wrote:
My favorite of this class of issues, was when I was working at a startup doing an
embedded product with Linux as the OS. We all had Linux machines for building/testing. I
was responsible for the Linux kernel at the company. I had someone come up to me and said
?the VM system is broken?my program works fine on my desktop but I keep getting page
faults on the embedded platform?. The conversation was something like this:
Me: So, what?s the program doing memory wise?
Him: I allocate an array of structures and I read them off of disk into the array.
Me: How big are the structures?
Him: About 4K
Me: How many are you allocating?
Him: 100,000
Me: Rolls eyes. That?s 4GB. We don?t have 4GB on the embedded platform. You don?t even
have that on your workstation.
Him: But it works there and not on the embedded platform.
Me: We don?t have paging turned on in the embedded platform.
Him: Can?t we turn it on? It makes my code so simple!
Me: No! Go allocate a *reasonable* sized array (say 100) and shuffle in from the disk
when the data you need isn?t there. You walk the data structure mostly sequentially
anyway, so there?s no reason to have the entire file in memory.
I still think a 3 tape sort should be a required early assignment (learn that not
everything can fit into memory at once).
TTFN - Guy
On Oct 30, 2015, at 2:16 PM, Noel Chiappa <jnc
at mercury.lcs.mit.edu> wrote:
...
Apparently nobody had ever told them that a number in memory is... just a
number. At which point it became clear that they needed to know a little
more about how a computer actually worked.
That reminds me of an exchange I had with a younger colleague who was writing a document
describing how to do incremental TCP checksum update correctly. He was tripping over the
issue of negative zero. I told him the answer simply was to subtract the complement with
end around borrow. He didn't believe me and ended up proving it by exhaustively
testing every case.
1:40 p.m.
New subject: "Demystification" is just as important as "Abstraction" in "Computer Science" (Was: Know any Fortran
On Fri, 30 Oct 2015, Guy Sotomayor wrote:
Him: I allocate an array of structures and I read them
off of disk into the array.
Me: How big are the structures?
Him: About 4K
Me: How many are you allocating?
Him: 100,000
Me: Rolls eyes. That?s 4GB. We don?t have 4GB on the embedded platform. You don?t even
have that on your workstation.
Him: But it works there and not on the embedded platform.
Me: We don?t have paging turned on in the embedded platform.
Him: Can?t we turn it on? It makes my code so simple!
. . .
I still think a 3 tape sort should be a required early assignment (learn that not
everything can fit into memory at once).
I struggled with that for 30 years.
Third semester programming students in Data Structures and Algorithms
classes still insisted on trying to load the whole file into memory.
Even in the simplistic example that I gave them of a business appending a
record to an already sorted file and then sorting it, when they should
have sorted just the appended records and then done a two file input, one
file output merge, with only two records in memory at a time.
I always assigned them a problem of writing a sort for a file/dataset too
big to fit into memory. They literally refused to understand that a
dataset could exist that would be too large to fit into memory.
A couple of students wrote a letter to the college
administration complaining about me and the department and our
"out-dated" curriculum, insisting that "in the real world, the proper
solution to a file too big for memory is to replace the computer with a
bigger one." I was actually called on the carpet to answer the charge!
That was trivially easy when I pointed out to the chancellor how much it
would COST to equip the student homework computer labs with computers
capable of holding a national telephone directory in memory.
Our industry is saturated with such refusal to understand. Consider
Microsoft's "throw hardware at the software problems" argument that if
their software runs too slowly on your computer, then that shows that YOUR
COMPUTER (and you) is inadequate, and you should get a faster computer.
--
Grumpy Ol' Fred cisin at xenosoft.com
2:14 p.m.
New subject: "Demystification" is just as important as "Abstraction" in "Computer Science" (Was: Know any Fortran
On 10/30/2015 01:40 PM, Fred Cisin wrote:
I always assigned them a problem of writing a sort for
a file/dataset
too big to fit into memory. They literally refused to understand
that a dataset could exist that would be too large to fit into
memory. A couple of students wrote a letter to the college
administration complaining about me and the department and our
"out-dated" curriculum, insisting that "in the real world, the proper
solution to a file too big for memory is to replace the computer with
a bigger one." I was actually called on the carpet to answer the
charge! That was trivially easy when I pointed out to the chancellor
how much it would COST to equip the student homework computer labs
with computers capable of holding a national telephone directory in
memory.
How times change. Most programmers of my era, would do it the "old" way
by habit--read in bits of the dataset. It took quite a bit of
relearning that mapping a file into a 48-bit address space and operating
on the data directly, letting the pager do the bookkeeping could render
what at first started out to be a fairly complicated job, very simple.
--Chuck
2:21 p.m.
New subject: "Demystification" is just as important as "Abstraction" in "Computer Science" (Was: Know any Fortran
From: "Fred Cisin" <cisin at xenosoft.com>
They literally refused to understand that a
dataset could exist that would be too large to fit into memory.
In the UK, the Home Secretary wants to force all ISPs to store and
keep (reasonably) easily searchable logs of all URLs accessed by all
their customers.
This, I think, is an example of "big data" ? possibly so big that backup
is impracticable.
When I did daily analysis of the urls accessed by the 2000 or so students
at a University*, that daily database was pushing what I could get into memory
on my desktop system (well specified) ? now imagine 20 million users for 365 days -
I'd be doubtful
that even todays mainframes and supercomputers could do that.
* When I questioned the legality (and morality) of doing this the answer was
"we believe it is legal" (and "stop complaining and keep digging").
Andy
2:37 p.m.
New subject: "Demystification" is just as important as "Abstraction" in "Computer Science" (Was: Know any Fortran
> They literally refused to understand that a
> dataset could exist that would be too large to fit into memory.
On Fri, 30 Oct
2015, ANDY HOLT wrote:
In the UK, the Home Secretary wants to force all ISPs
to store and
keep (reasonably) easily searchable logs of all URLs accessed by all
their customers.
This, I think, is an example of "big data" ? possibly so big that backup
is impracticable.
When I did daily analysis of the urls accessed by the 2000 or so students
at a University*, that daily database was pushing what I could get into memory
on my desktop system (well specified) ? now imagine 20 million users for 365 days -
I'd be doubtful
that even todays mainframes and supercomputers could do that.
* When I questioned the legality (and morality) of doing this the answer was
"we believe it is legal" (and "stop complaining and keep digging").
That would have been a much better example to bring up with the
Chancellor, since the Peralta Colleges administrators would LOVE
to keep track of all student computer usage!
He would have thought that it was a trivial requirement until walking him
through the arithmetic.
At one time, they had a "Chief Information Officer" who was joining
[temporarily?] various Yahoo email groups for the sole purpose of access
to archives to look for faculty posting "inappropriate" content (any
defamation of Peralta administration!) Yes, John Wxxxxxxx, eat shit and
die.
2:43 p.m.
New subject: "Demystification" is just as important as "Abstraction" in "Computer Science" (Was: Know any Fortran
-----Original Message-----
From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of ANDY
HOLT
Sent: 30 October 2015 21:22
To: General Discussion: On-Topic and Off-Topic Posts
<cctalk at classiccmp.org>
Subject: Re: "Demystification" is just as important as "Abstraction"
in
"Computer Science" (Was: Know any Fortran
From: "Fred Cisin" <cisin at xenosoft.com>
About a year ago, I was asked by my employer to give them a list of the URLs accessed by
one member of staff. I sent them a list of the URL's that I had accessed.
It was totally useless, mainly because every page these day as is a composite. Add to that
allowing Outlook to download the content of an HTML e-mail ..
.. its almost a totally pointless exercise....
They literally refused to understand that a
dataset could exist that
would be too large to fit into memory.
In the UK, the Home Secretary wants to force all ISPs to store and keep
(reasonably) easily searchable logs of all URLs accessed by all their customers.
This, I think, is an example of "big data" ? possibly so big that backup is
impracticable.
When I did daily analysis of the urls accessed by the 2000 or so students at a
University*, that daily database was pushing what I could get into memory
on my desktop system (well specified) ? now imagine 20 million users for 365
days - I'd be doubtful that even todays mainframes and supercomputers
could do that.
* When I questioned the legality (and morality) of
doing this the answer was
"we believe it is legal" (and "stop complaining and keep digging").
Andy
3:39 p.m.
New subject: "Demystification" is just as important as "Abstraction" in "Computer Science" (Was: Know any Fortran
>> They literally refused to understand that a
dataset could exist that
>> would be too large to fit into memory.
On Fri, 30 Oct 2015, Dave Wade
wrote:
About a year ago, I was asked by my employer to give
them a list of the
URLs accessed by one member of staff. I sent them a list of the URL's
that I had accessed.
Or you could "comply" by listing only URLs that were typed in.
"He typed in GOOGLE.COM, and spent hours before he typed any other URLS.
No idea what he did on that site; he might have filled in fields, or
clicked on things. How would you feel about installing a camera to look
over his shoulder?" (let them acknowledge what they are really doing.)
4:05 p.m.
New subject: "Demystification" is just as important as "Abstraction" in "Computer Science" (Was: Know any Fortran
-----Original Message-----
From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Fred
Cisin
Sent: 30 October 2015 22:40
To: General Discussion: On-Topic and Off-Topic Posts
<cctalk at classiccmp.org>
Subject: RE: "Demystification" is just as important as "Abstraction"
in
"Computer Science" (Was: Know any Fortran
I couldn't figure out how to work that out without installing a key logger
on the users PC....
>> They literally refused to understand that
a dataset could exist that
>> would be too large to fit into memory.
On Fri, 30 Oct 2015, Dave Wade
wrote:
About a year ago, I was asked by my employer to
give them a list of
the URLs accessed by one member of staff. I sent them a list of the
URL's that I had accessed.
Or you could "comply" by listing only URLs that were typed in. "He typed in GOOGLE.COM, and spent hours before
he typed any other
URLS.
No idea what he did on that site; he might have filled in fields, or
clicked on
things. How would you feel about installing a camera
to look over his
shoulder?" (let them acknowledge what they are really doing.)
5:52 p.m.
New subject: "Demystification" is just as important as "Abstraction" in "Computer Science" (Was: Know any Fortran
On Oct 30, 2015, at 5:21 PM, ANDY HOLT <andy.holt
at tesco.net> wrote:
From: "Fred Cisin" <cisin at xenosoft.com>
Orwell's inspiration must have come from nearby.
Meanwhile, Tor to the rescue.
paul
They literally refused to understand that a
dataset could exist that would be too large to fit into memory.
In the UK, the Home Secretary wants to force all ISPs to store and
keep (reasonably) easily searchable logs of all URLs accessed by all
their customers. 
3:08 p.m.
"I still think a 3 tape sort should be a required early assignment (learn
that not everything can fit into memory at once).
TTFN - Guy"
Second.
Lee C.
On Fri, Oct 30, 2015 at 12:15 PM, Guy Sotomayor <ggs at shiresoft.com> wrote:
writing a
document describing how to do incremental TCP checksum update
correctly. He was tripping over the issue of negative zero. I told him
the answer simply was to subtract the complement with end around borrow.
He didn't believe me and ended up proving it by exhaustively testing every
case.
My favorite of this class of issues, was when I was working at a startup
doing an embedded product with Linux as the OS. We all had Linux machines
for building/testing. I was responsible for the Linux kernel at the
company. I had someone come up to me and said ?the VM system is broken?my
program works fine on my desktop but I keep getting page faults on the
embedded platform?. The conversation was something like this:
Me: So, what?s the program doing memory wise?
Him: I allocate an array of structures and I read them off of disk into
the array.
Me: How big are the structures?
Him: About 4K
Me: How many are you allocating?
Him: 100,000
Me: Rolls eyes. That?s 4GB. We don?t have 4GB on the embedded platform.
You don?t even have that on your workstation.
Him: But it works there and not on the embedded platform.
Me: We don?t have paging turned on in the embedded platform.
Him: Can?t we turn it on? It makes my code so simple!
Me: No! Go allocate a *reasonable* sized array (say 100) and shuffle in
from the disk when the data you need isn?t there. You walk the data
structure mostly sequentially anyway, so there?s no reason to have the
entire file in memory.
I still think a 3 tape sort should be a required early assignment (learn
that not everything can fit into memory at once).
TTFN - Guy
--
Lee Courtney
+1-650-704-3934 cell
On Oct 30, 2015, at 11:35 AM, Paul Koning
<paulkoning at comcast.net>
wrote:
> On Oct 30, 2015, at 2:16 PM, Noel Chiappa <jnc at mercury.lcs.mit.edu>
wrote:
>
> ...
> Apparently nobody had ever told them that a number in memory is... just
a
number.
At which point it became clear that they needed to know a little
more about how a computer actually worked.
That reminds me of an exchange I had with a younger colleague who was
3:13 p.m.
> "I still think a 3 tape sort should be a
required early assignment
> (learn that not everything can fit into memory at once).
On Fri, 30 Oct 2015,
Lee Courtney wrote:
Second.
It's hard to get students to care about what you are saying when you
mention tapes. You need to explain the algorithms using files.
Even though CD/DVD is a sequential storage system.
6:47 p.m.
Noel Chiappa wrote:
It's worth teaching a bit of machine/assembler
language, so that students
understand how computers _actually work_, underneath.
[example of students who wrote octal math for CLU]
Back in 1983 I needed an assembler for the Motorola 6809, so I wrote one
in Lisp on a TRS-80 Model I. I know very well how things really work (I
design microprocessors at the transistor level) and yet I created a hex
math package in Lisp just like in your example. Sure it was orders of
magnitude slower than using the language's native math and doing all the
conversions in the i/o routines, but it was very elegant and was fast
enough for the sizes of the programs we needed to assemble.
So while I insist on knowing all the details, I also enjoy tools that
don't depend on me knowing.
-- Jecel
3385
days inactive
3386
days old
14 comments
10 participants
participants (10)
-
andy.holt＠tesco.net -
cclist＠sydex.com -
cisin＠xenosoft.com -
dave.g4ugm＠gmail.com -
ggs＠shiresoft.com -
jecel＠merlintec.com -
jnc＠mercury.lcs.mit.edu -
leec2124＠gmail.com -
paulkoning＠comcast.net -
toby＠telegraphics.com.au