🚀 go-pugleaf

AbsoluteBurn@usa.net wrote:
> 
> I was thinking of starting a new Borland C++ website.

Basically to collect information at one place and providing easy access
to it is a great idea.

But there are some questions one has to think about very thoroughly
before enthusiastically starting a project that might end up with the
same problems others already couldn�t solve.

A) I'm a bit worried about the easy access part. That's one of the
biggest problems in any library. It doesn't make any sense to store all
the information and all the solutions if one can't find it. This
newsgroup proofs that a lot of people have trouble finding information.
There are so many questions posted in the newsgroup that had been
answered quicker and sometimes more competent and comprehensive if they
just had a look at the appropriate help file.
So, what�s the problem? I�m afraid some people prefer to let others do
the work they�re supposed to do. But I guess most of them have really
tried hard and just couldn�t find it.
Example: Over the last few months there have been about ten articles
posted asking about setting text in and retrieving text from a dialog
edit control. If you open the help file it�s exactly three steps until
you are at the right point. There�s almost no way of missing the answer.
So how come they don�t find it? Don�t they know that the help file
exists? Don�t they know how the help file is structured? Does it mean
they�re looking for an apple but they don�t know that it grows on trees?
Or if they know, maybe the don�t know that the tree simply is called
apple tree? Or, most likely, is the �search engine� to stupid or to
confusing to support them properly?

B) Let�s assume we�ve found what we�ve been looking for. Then another
issue is the selection of explanations and code examples. What�s a good
example, what�s a bad example and who�s going to decide that? Borland�s
documentation for instance is a good example of a fine selection of bad
examples. This collection of classes A, classes B, classes X, classes
Derived, functions f() handles the issues sometimes so theoretically and
confusing that it�s no help at all. Who�s supposed to know after reading
two pages of source that doesn�t do anything useful but trying to do
everything possible whether class Derived had something to do with class
X or class A or not?
Have a look at Borland�s documentation concerning dynamic_cast and then
have a look at http://www.acm.org/crossroads/xrds3-1/ovp3-1.html. I
think in 99 of 100 cases explaining something using real live issues
like bank accounts is much easier to understand than Borland�s
dynamic_cast<T>(*p). And it spares you unnecessary explanations like:
�if class B is derived from class A� or �remember that class B was
derived from class A�. It�s obvious that a savings account is a special
case of bank account and therefore derived from bank account and you
don�t have to be remembered even after going through ten pages of
source.
Coming back to the point: How can one make sure that what you�ve found
is really helpful? How to make sure that not just only a chosen few
understand the meaning? How to make sure that an example explains
everything thoroughly without going to extremes and/or confusing people.
The explanation of dynamic_cast at
http://www.acm.org/crossroads/xrds3-1/ovp3-1.html for instance is a good
example of a well done explanation that just stops at the point in front
of the gate to �C++ Wonderland�, it�s a pity.

Anyway, I guess most of us would more than appreciate a �tool� like that
and I would really like to assist.

J�rg