Re: Try Finally...

From: VBDis (vbdis_at_aol.com)
Date: 10/28/04 

Date: 28 Oct 2004 07:59:18 GMT

Im Artikel <Enrfd.3799$Km6.86971@news4.e.nsc.no>, "Frank de Groot"
<franciad@online.no> schreibt:

>I don't know much about exceptions, but it is plain that any reserved word
>in a language should not be implemented as 40 kb in some unit..

It's just a welcome feature of any programming language, when a complex
operation can be achieved with simple and short syntax. Isn't this just the
reason for using high level programming languages, libraries and components at
all, to reduce the efforts required to make things work?

It's another thing when the according library code is not thoroughly
documented. This shortage does no harm as long as the description of the usage
is sufficient, apart from nasty implementation details. But the current
discussion goes beyond the user interface, down into implementation details and
into the possibly not clearly enough stated model of operation. Here we may ask
whether the same task could be achieved with less code, but then we should not
wonder when we find out that an implementation of the described user interface
really requires that amount of code, and that modifications will break the
described behaviour.

A reduction of code size typically means a reduction of existing features. In
some cases it will be acceptable to remove unused features, and to use e.g. API
Windows and Buttons instead of Delphi Forms and Controls. But usually such
questions are asked by people that don't have to bill their time, or which
didn't yet realize how much time it can take to reduce the size of some code.
Laura is a professional, and so I don't ask wether it's worth the time spent
with a redesign of the Delphi RTL. My only interest here is to learn more about
SEH, and what can be changed when what restrictions or different behaviour of
the resulting code is acceptable. In so far it already has been worth the time
that I've spent with this thread, and the time I possibly will continue to
spend with it.

DoDi

Quantcast