On Friday 02 October 2009 05:10:06 am Eleanor McHugh wrote:
The best teacher of “the right way” is experience, and experience
mostly comes from doing things the wrong way and then having to clear
up the mess at some appreciable cost to yourself.
That is true for the species in general. It is not true for the
individual.
For example, we, as a species, have enough experience to know that
shooting
yourself in the foot will probably hurt, and may even require that foot
to be
amputated. It would be ridiculous to expect every individual to do that.
Having to clear up the mess is helpful, except for the situation where
you
don’t realize what a mess you’re in.
I suspect most of
the problems big companies have (and not just with coders) is that
when a person works on one small problem and then passes their output
along to a coworker in an equally myopic position, the producers of
mess are rarely if ever penalised.
That is a large part of it. And having responsibility for the mess would
mean
someone would try that much harder to avoid the mess in the first place.
But the key word there is “avoid”.
Some of these people are the people who will grow up to use Visual
Basic, and
enable “on error resume next”.
I guess you probably don’t approve of rescue nil in ruby either lol
In production? Nope.
In development? Be careful, because it might survive till production.
I’d much
rather replace the troublesome code with an explicit stub than have
something
that half-works, maybe, this time.
It’s actually something I’m starting to dislike about shell scripting:
#!/bin/sh
source config/variables.
rm -rf ./$BACKUP_DIR
cp -a $FILES ./$BACKUP_DIR
…
Obivously, BACKUP_DIR is assumed to be defined in config/variables. But
if that
file doesn’t exist, “source” will raise an error – but the next line
will
proceed. If the file does exist, but doesn’t define BACKUP_DIR, the
result is
the same – in either case, BACKUP_DIR will be expanded to an empty
string,
resulting in:
rm -rf ./
It’s a contrived example, but not too far off from reality. And I don’t
think
anyone should have to see their files go away to learn to do proper
error
checking in their shell scripts. It can be as easy as adding a && to the
end
of each line except the last – and there’s probably some even easier
flag to
set…
Same with Perl and “use strict; use warnings;” – I had to learn that
those
exist first before I could use them properly. If I was only to learn
from
experience, I’d probably have written years worth of programs in which
misspelling a variable didn’t raise an error, but instead gave me a null
value.
