From: Stu <stu@...>
Date: 2011-04-04T18:28:21+09:00
Subject: Re: functional paradigm taking over

is this a good book to read to grok functional programming,
http://mitpress.mit.edu/sicp/full-text/book/book.html

I have done some lisp but interested in gaining the skillset with ruby,

any nice thick comp sci books for recommendation?

On Mon, Apr 4, 2011 at 4:05 AM, Mike Stephens <rubfor@recitel.net> wrote:
> This thread has been touching upon three issues - functional languages
> as a way of expressing problems, Excel as a language environment, and
> the programmer.
>
> Today, relational databases are widely accepted. However Ted Codd did
> not invent the relational model because he thought relations were neat.
> If you look at what he said at the time, his motivation was to make
> corporate data more accessible to non-programmers. Forty years ago he
> thought there were less programmers than there were programs to be
> written. Arguably the same is true today.
>
> When I was a boy, I would see IDMS recordsets like this:
>
> 200
> 300
> 150
> 200
> 230
> etc
>
> What this represented was
>
> 2011/Q1 Sales = £200
> 2011/Q2 Sales = £300
> 2011/Q2 Sales = £150
>
> Worse still you would have:
>
> 12
> 200
> 27
> 300
> etc
>
> which meant:
> 2011/Q1 Sales = £200
> 2011/Q1 Units = 12
> 2011/Q2 Sales = £300
> 2011/Q2 Units = 27
> etc
>
> Tedd said you must use relations. Relations are unordered so you can't
> hide information in the order of data. Also you must have a field name
> for each field (and it should be on the same domain, so Sales shouldn't
> be in the same 'column' as Units).
>
> You would therefore have to have:
>
> Period Sales Units
> 2011/Q1 200   12
> 2011/Q2 300   27
> etc
>
> Whilst- as we all know - that typically doesn't tell you everything you
> need to know when you come across a recordset, it makes it much easier
> for someone not deeply involved in the system concerned, to work out
> what is going on.
>
> Put it another way, using relations makes it less likely you will make a
> mistake because you didn't know the subtleties of the behaviour of the
> programs generating that data.
>
> Then, relations are amenable to first order predicate calculus. This
> resticts the complexity of the operations you need to carry out.
>
> When you consider the merits of functional languages, I believe you
> should take into account not only the elegance with which you can
> express what you want to do, or how amenable it is to eg parallel
> processing. You should also consider whether it strikes a better balance
> between such issues and accessibility. Does all your terseness and
> 'elegance' give you a justifiable advantage over something simpler which
> could be more easily handled by a person of less capability?
>
> --
> Posted via http://www.ruby-forum.com/.
>