Re: [9fans] Simple

Re: [9fans] Simple

[Laura Creighton]

>simple?  that's hard.  you either understand it or you don't.
>boyd

No.  It *is* teachable.  You have to hold somebody's feet to the fire
and, line by line, say ``what does that piece of code do.  Now. Is that
something that we need to do.  And is that something we need to be doing
*here*.''  But the only reason I know how to do it is because a long
time ago Geoff Collyer and Henry Spencer (hi Geoff! thank you by the way)
held *my* feet to the fire and did the same thing.  A surprising amount of
bad code is in programs because the person who wrote it does not know
what the code does, and does not want to admit ignorance, even to themselves.
This is programming by sympathetic magic.  These charms and those chants
somehow made that code work over there, so I had better put some charms and
chants in my code as well.

I have spent a good part of today trying to understand why boyd's e acute
came to me as an e and not as an i.  After wading around in MIME doc for
a while I have much more sympathy for the sympathetic magic approach. I
could understand this stuff, but boy oh boy the temptation to try to get
away with using it without understanding it is enormous.

Re: [9fans] Simple

[Boyd Roberts]

From: Laura Creighton <lac@cd.chalmers.se>
> 
> No.  It *is* teachable.

yes but you have to be able to _learn_ it.  some people can't.

eg. the '1000 line shell script brigade' etc, etc...

> I could understand this stuff, but boy oh boy the temptation
> to try to get away with using it without understanding it 
> is enormous.

i believe most implementations get away with it.  i don't fully
understand it, but i do understand it is not to be attempted if
a) it can be avoided or b) you don't understand it _completely_.

even relatively simple stuff like 822 groups microsoft couldn't
get right (not that i think groups are a good thing, but i do
like them to break their user agents).

Re: [9fans] Simple

[kim kubik]

-----Original Message-----
From: Laura Creighton <lac@cd.chalmers.se>
To: 9fans@cse.psu.edu <9fans@cse.psu.edu>
Cc: lac@cd.chalmers.se <lac@cd.chalmers.se>
Date: Thursday, November 23, 2000 1:45 PM
Subject: [9fans] Simple


>>Yeah, but doesn't the S in SMTP mean simple?
>>-rob
>
>Next time you write a book, do you think you could devote a chapter on
>what is simple?

May, 1986 interview with Dick Haight (then Supervisor of
Technical Staff at Bell Labs) in "Unix Review":

Questioned about the move to GUIs and how it affects
a Unix that can only offer shells:

  "it messes up what has always been a beautifully
   simple programming system.  UNIX has only a small
   number of system calls and is amazingly consistent
   from on layer to the next.  Ken Thompson took great
   care in writing a lot of those inner layers.  You
   know, in the early days, it was an aesthetic experience
   to get deep down in the system."

and a bit later:

  "I still remember the first time I read Ken Thompson's
   recursive descent assembly language parsing.  It was
   sort of like Paul's vision on the road to Damascus -
   in programmer's terms, of course."

Sounds like "simple" to me .  .  .

Best part of interview, though, was, "I remember that on
one unannounced visit to the attic UNIX lab in Murray Hill,
some of these three-piece-suit guys found Ken Thompson in
the carcass of a PDP-11 - like he was under a car working
on the oil pan or something.  The culture shock was more
than those management guys could stand.  I think that
visit set PWB back a month or two."

There's some other interesting history about putting JCL
error (IBM systems) to get top priority on a resubmitted
job and send to files different machines by 'faking' them 
to be line printers. . .

 - k  

Re: [9fans] Simple

[Douglas A. Gwyn]

nigel@9fs.org wrote:

> Einstein understood it, apparently, when he said
> "Everything should be made as simple as possible,
> but not one bit simpler."

That's not an exact quote, but close enough..

> I can't comment if this applied to his theories.

What Einstein was addressing was physical theory,
and the point was that an optimum theory would
take into account *all* relevant facts while
explaining them in the most direct way possible.

It doesn't have much to do with software design,
but if you tried to stretch the analogy it would
mean that software should meet *all* specified
requirements in the most direct way possible.

Re: [9fans] Simple

[nigel]

[-- Attachment #1: Type: text/plain, Size: 190 bytes --]

Einstein understood it, apparently, when he said

"Everything should be made as simple as possible,
but not one bit simpler."

I can't comment if this applied to his theories.





[-- Attachment #2: Type: message/rfc822, Size: 1672 bytes --]

From: "Boyd Roberts" <boyd@planete.net>
To: <9fans@cse.psu.edu>
Subject: Re: [9fans] Simple
Date: Thu, 23 Nov 2000 22:52:34 +0100
Message-ID: <029d01c05597$aff3ff80$0ab9c6d4@cybercable.fr>

simple?  that's hard.  you either understand it or you don't.

Re: [9fans] Simple

[Boyd Roberts]

simple?  that's hard.  you either understand it or you don't.

[9fans] Simple

[Laura Creighton]

>Yeah, but doesn't the S in SMTP mean simple?
>
>-rob

Next time you write a book, do you think you could devote a chapter on
what is simple?  It is badly not understood even by people who are
trying to write good code.  When I tell people, ok, now it works.
Now go away and strip out things until this is the simplest and most
elegant piece of code that you can write that works, people look at
me as if I had sprouted an extra head or something.