Re: pl/sql code maintenance

Malcolm Dew-Jones wrote:
>
> Jim Kennedy (kennedy-down_with_spammers_at_attbi.com) wrote:
> : Reminds me of 3 stories.
>
> There are two sides to everything, and a few side thoughts comes to mind
>

[shipping bad code on damaged discs story snipped]
>
> : Another company (made tape backup software) would put in wait states in
> : patches so things would take longer and longer to backup. Then they would
> : market the next major upgrade as a speed improvement. (which you had to buy
> : to get)
>
> (no other side to this story!)

Also a very old story. It was done in hardware first. The "upgrade" to a faster printer required cutting one wire.
>
> : A large computer company (known by three letters), had a requirement to have
> : a timesharing system where the performance of a benchmark would be the same
> : for 1 user or if all 30 users were on the system at the same time.
> : Engineering met this marketing requirement. They put in wait states in
> : inverse proportion to the number of users.
>
> This one sounds like a classic "dilbert".
>
> The key words here are "marketing requirement". The customer or
> management insists on impossible requirements and so the technical people
> find ways to keep them happy.

Actually there is a good human factors reason for this. Presented with two systems like this (note, response times are fictional but not absurd):

1. Very fast response (micro seconds) most of the time, but occasional delays of up to 10 seconds.
2. very consistent response between 1-2seconds, and no delays.

Users consider system B to be more reliable and responsive. Consistancy is actually very important to systems with human interfaces.

--
Ed Prochak
Magic Interface, Ltd.
440-498-3700(office)
Computer consulting, database and web services.