N
Nick Keighley
propably not. I'm sure he sincerly believes he's right
here I have to disagree. There is no way you can know this.
Unless you have some deep insight into bill's psychology
I don't see how *anyone* (at least on comp.lang.c) can
work out what would help bill. For this reason I didn't
contradict you when you suggested examining the data structures and
program flow using a debugger. Normally I don't see the point
(yep, we disagreed on this before); but in bills' case who then
hell can tell what *might* help. May be he'll play around in gdb
and click! the light goes on. I must admit the output from raw
gdb looks a bit messy (and bill confuses easily). Mightn't a gui
based one be better for him? ddd actually draws pictures of data
structures.
He just seems to lack some basic machinary or insight into
how imperetive programming languages work. Maybe he should
learn assembler. Maybe he should learn prolog. I just don't know.
*I* think (and I have nearly as much doubt about this as your
debugger-solves-everything approach), he lacks the ability to plainly
state what he is trying to do. I've tried english, pseudo-code
and even assembler (when I was starting some insight into the
hardware was fruitful- but people vary). if you can't say
what you want how can you ever know you've acheived it? I think
some people are just mentally unequipped to program. Bill may
be one of them.
ah. I've spent *some* time with trainees- and not recently
he certainly needs *something* to get him out of his current
"methodology".
"I wonder if a for-loop would be useful here". I've floundered
but never to that extent!
I've suggested paper execution (which I suppose is a manual way
of single stepping a debugger).
<snip>
--
Nick Keighley
"Perilous to us all are the devices of an art deeper than we possess
ourselves."
Gandalf The Grey (discussing Windows NT)
The point is that it would.
here I have to disagree. There is no way you can know this.
Unless you have some deep insight into bill's psychology
I don't see how *anyone* (at least on comp.lang.c) can
work out what would help bill. For this reason I didn't
contradict you when you suggested examining the data structures and
program flow using a debugger. Normally I don't see the point
(yep, we disagreed on this before); but in bills' case who then
hell can tell what *might* help. May be he'll play around in gdb
and click! the light goes on. I must admit the output from raw
gdb looks a bit messy (and bill confuses easily). Mightn't a gui
based one be better for him? ddd actually draws pictures of data
structures.
He just seems to lack some basic machinary or insight into
how imperetive programming languages work. Maybe he should
learn assembler. Maybe he should learn prolog. I just don't know.
*I* think (and I have nearly as much doubt about this as your
debugger-solves-everything approach), he lacks the ability to plainly
state what he is trying to do. I've tried english, pseudo-code
and even assembler (when I was starting some insight into the
hardware was fruitful- but people vary). if you can't say
what you want how can you ever know you've acheived it? I think
some people are just mentally unequipped to program. Bill may
be one of them.
You might not see it. But you are looking
from a different, possibly equally valid, angle. I have spent a lot of
time with trainees
ah. I've spent *some* time with trainees- and not recently
and I have seen people almost as stuck as Bill.
wow
Best
way is to get them into the memory and SEE things as the program
steps. To see the string. To see the chars. To see the end of string
marker.
And if you think this advice is trolling then I am surprised.
In some cases I take that stance. For example when someone assures me
they can debug 50,000 lines of foreign c code quicker from a printout
than from using an industry strength debugger. Those kind of things that
only get spouted in clc. it is nonsense. Can someone in the world? I
have no doubt someone can. Would we recommend that to someone like Bill
because some Indian Guru can? No we would not. Would we guess that 99%
can because Heathfield once spotted a bug on page one 10 years ago? No
we would not. We would advise sensible procedures to approach the
problem.
My position here is that its the old seed versus crops thing with
Bill. Get him to teach himself the basic types. Feeling them with a
debugger is a great way. He seems to be as clueless as 6 months ago and
need to get his hands dirty.
he certainly needs *something* to get him out of his current
"methodology".
"I wonder if a for-loop would be useful here". I've floundered
but never to that extent!
I've suggested paper execution (which I suppose is a manual way
of single stepping a debugger).
Pasting in solutions for him is not helping
him -
agreed
if anything it is hindering him since he then mistakenly thinks he
knows why something is working. He doesn't. He needs to follow the
tutorial and learn how to examine his own programs for now. More
complicated things can follow.
<snip>
--
Nick Keighley
"Perilous to us all are the devices of an art deeper than we possess
ourselves."
Gandalf The Grey (discussing Windows NT)