On 28/01/13 08:55, Joe Zeff wrote:
On 01/27/2013 01:00 PM, David WE Roberts wrote:
I still don't understand why the latest threads want to snuggle up next to
the negative scores instead of the positive scores.
Sorting on several fields can be quite tricky. I remember, once, many
many years ago, changing a very slow bubble sort of a list of customers
to a shell sort that took 5% of the time. The data entry person
complained because the sort wasn't "stable." That is, if there were
multiple customers with the same last name (what I was asked to sort
on) the new list didn't keep the customer numbers in correct order,
something I'd not been asked to do.[1]
If you're not asked for a stable sort, you can't exactly be criticised for not
providing a stable sort. Much.
In practice, "sorting should be stable" should be one of those no-brainers for
user-facing applications. A bit like, "what do you mean, you wanted the Save command to
actually write the data to disk? Nobody specified that in the spec!"
(Aside: or an oven that can actually heat up food. My oven has *three* off
settings. There's OFF, and then there's LIGHT, and then there's FAN. To
actually turn the heating element on, you need to set it to FAN ASSIST. I can't
tell you how many times I've turned the oven on and wondered why it wasn't
warming up because I had it on FAN instead of FAN ASSIST.)
[1]She'd asked for this because with the old routine she could have
printed out the list on index cards and hand sorted them faster. I
never said it, but I couldn't help thinking that some people are never
satisfied, even if they get exactly what they asked for.
What they ask for is not necessarily what they were expecting, because they
thought that it went without saying. Or they didn't realise they needed it
until it was gone. This is why all specs should always go through at least two
iterations.
The correct answer to "But this is what you asked for" is often "Yes, and now that
you've delivered it, I realise that it's not what I need".
--
Steven
_______________________________________________
Pan-users mailing list
Pan-users@nongnu.org
https://lists.nongnu.org/mailman/listinfo/pan-users
