Subj : Re: Software Job Market Myths
To   : comp.programming,comp.software-eng
From : rem642b
Date : Mon Aug 15 2005 01:14 pm

> From: Alan Balmer <albal...@att.net>
> Programming is not normally something that needs to be done
> reflexively or even quickly (in terms of the length of an interview.)
> I'd rather have a programmer who has an excellent solution in a day
> than a sorta good one in 5 minutes.

If I were the boss, I'd like to know within 5 minutes whether he even
understands the question, and within a half hour or so what various
ideas he has for plans toward actually solving the problem. If he
doesn't understand the question, or his ideas for solution are totlally
inappropriate, then I'm not going to pay for 8 hours of his work just
to throw it all away because it's moot to the problem at hand. I don't
expect he's decided on a plan within a half hour, but at least his
various options should mostly sound plausable.

For example, suppose the problem is that our mail servers have been
getting flooded with spam, and we've made a policy decision that we are
going to start telephoning the major problem ISPs with demands they
stop sending us spam or we'll configure our SMTP servers to block all
connections from them. What software will we need?

Suppose the person says: The the only thing in spam that can be trusted
is the IP number of the SMTP client that connected to our SMTP server.
So we need to collect those IP numbers, either directly from the log of
our SMTP server, or from the Received: line that our server attaches to
each incoming message. So we need either tracking between server and
messages that are later recognized as spam, or parsing of headers to
recover the IP number. Then we need a map from IP numbers to
responsible ISPs. This is provided by WHOIS servers, so we need
software to connect to WHOIS servers and perform searches and collect
the records and parse it for the relevant information. Then we need a
database to store all the collected data and check whether we already
have the data for a new IP number or need to go get it from scratch.
Then we need software to audit our database to see which ISPs have been
causing the most problem so that we can concentrate our efforts there.

I'd classify that as a workable starting plan. I'd then make the
administrative decision whether to do SMTP server tracking or header
parsing, and check with other employees who might have expertise in
some of the needed areas to speed along the project so this one person
doesn't have to code everything from scratch.

By the way, unless you can split this task among lots of employees each
coding one part of the overall task, it's going to take a lot longer
than a day or even a week to get the whole job done properly.

.