Re: Dynamicly declaring arrays (was: Time to join the fold?)

[Matthew Heaney <matthew_heaney@acm.org>]

dennison@telepath.com writes:

> In article <m3aeyz8mw1.fsf@mheaney.ni.net>,
>   Matthew Heaney <matthew_heaney@acm.org> wrote:
> > dennison@telepath.com writes:
> >
> > > In a severe jam a declare block can be used to get a localized
> > > constant string, or one of the ada.strings packages can be used.
> >
> > Rather odd to describe using a declare block to declare a string object
> > only in a "severe jam," since this is the normal, intended usage!
> >
> 
> True, that is what they are there for. But my style is to avoid declare
> blocks if I can help it. Typicaly I'd try to break off a proper subroutine
> first. When I do use declare blocks, its usually just to provide a localized
> exception handler (and in that case of course the "declare" isn't nessecary).
> 
> I'm not entirely sure I can defend this aversion to declare blocks in a
> rational way, so if anyone has a rationale as to why its perfecly acceptable
> to have declare blocks liberally sprinkled throught their code, I'm all ears.


The purpose of a declare block is to localize an object declaration, so you
can minimize the distance between the declaration of object and its
actual use.

The purpose of a subprogram is algorithmic abstraction.  A declare block
is not an abstraction of an algorithm.

I do "liberally sprinkle my code" with declare blocks, especially for
declaring the result of a function that returns an unconstrained array
(i.e. string).  This localizes the object, and avoids an assignment of a
string value to a string object, so there won't be any length mismatch
problems.

The point of this thread is about how to effect an ada mind-set.  The
liberal use of declare blocks is a common Ada idiom.