I have learnt C second language 26 years ago and have constantly used it
since despite this I am currently instituting a NO C policy at work.
Most (all) of our code does not need the performance of C and generally
the code is not optimal anyway and leaving the decisions to an optimised
scripting language is quite probably going to run faster than poor C
code written in a hurry.

If we work on numbers.  My code generally runs on a 2% error rate (all
right I am making 2% up but the number is not important...).  I rewrote
an application of 2,300 lines of C code (not a comment in sight) into
230 lines of perl code (with lots of comments).  Now with my 2% error
rate I had 5 errors in my code (about right).  Now the 2,300 lines would
have had 50 errors given the same error rate.  Don't use C because you
will create more code, more code equals more bugs.

The moral is use the correct language for the job.  Performance is just
a few thousand dollars in hardware away.  Take $40K as your base (for
example) and a server of $5K you need to save about 7 weeks work to pay
for it. It is not a lot of time with 10 times the number of errors in
your code.  Bugs cost the time in development, correcting bugs in
production code is far far worse.

Sounds like a talk coming one here...

Ken Foskey

-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Reply via email to