Re: Ada packages and/on web pages (was Re: GPGME (was Re: GnuPG binding?))

[Georg Bauhaus <sb463ba@d2-hrz.uni-duisburg.de>]

Preben Randhol <randhol+abuse@pvv.org> wrote:

(This text is drifting towards "writing Ada programs compared to
writing web pages", among other things.)

: not sure I understand what you mean with "Google establishes useful
: links between".

If the search for "A +B" gives results r(1..4), then there
might be a logical link between the r(k)s, because they match the
same search pattern.

: But have you considered that Google is your weak link?

(My link to interesting information? Sure, and it could have some
influence on business if Google and relatives go out of business,
for political or legal reasons for example.
(All the more we will need good links.))
 
:> However, compare setting up a software related web site to organizing
:> your latest Ada project into packages.
:> 
:> ...
:> - You can have a number of tasks.
: 
: What is number of tasks?

task A is
task B is

makes 2 tasks. In your web pages, you might have 
task "About us",
task "the famous software library"
task "the CI CSSs"
in the sense that the things to be done (written, set up) can be grouped
and treated separatedly. Though not necessarily without "rendezvous".
In fact, you could employ separate compilation ;-)

: But also you would have to set up some of these:
: 
:   Bugtracker :   Mailinglist :   CVS

In order to be competition for sourceforge, yes.
Plus if I think that for example a mailinglist is better than
group reply with permission. (Don't get me wrong, I have had
fun setting up and using a SmartList.)

Plus if I wanted to copy someone's machanisms. For example, I prefer
PRCS to CVS. At one time, we have been able to provide access to
a development system (non-intranet) via SSH, which has been much more
fun and a lot more flexible than anonymous CVS. I can do it if I
have my own Ada component library server.

 
:> Now how do you set up coupling between your modules? Whichever way,
:> will you think about the links between the program parts?
: 
: What do you mean?

See the posting by Robert Eachus on 16th of June, for coupling.
Say you have three sets of packages in your program,
and you have three subdirectories (e.g.) in your "web space",
by analogy.
Chances are that these three items have internal and external
relations, in both cases.
(calling, type use, instantiations, imports, ... in the Ada case,
links, quotes, styles, illustrations, ... in the web case.)
For example, a unit might depend on another unit.
You add a with clause.
A web page might assume required knowledge of subject X.
You add a link to another page (in another subdirectory, e.g.)
introducing subject X.

All three are also hopefully centered around some common topic.
You can have coherent Ada packages, and coherent web pages
or subdirectories thereof.
Meyer in OOSC2, Ch 22:
"All the features of a class must pertain to a single, well-identified
abstraction." (I think this is recommended for clusters, too.)
The same should be true of web pages, I think, and I always find myself
distracted by ads, links to other things that they think I should
be interested in, but am not, and more. If you, as author of a web page,
are allowed to follow the "single topic" principle, good.
The possibility is guaranteed if you are your own "web masters",
because there are no defaults and requirements other than your
own (lawful) ones.

:> Are there external programs that will play a role?
: 
: What has all this got to do with sourceforge f.ex? You can set up your
: own homepage with just the links you want etc...

Uhm, sourceforge has been suggested, iirc, to circumvent setting
up pages and links for download, and taking care of them. That is,
not having to set up your own homepage, download area, etc.

I wanted to say, mainly, that it might be better if people make an
effort at both producing good software, and also at producing an
accessible web page that sticks to the subject, which is their software,
and to really related subjects. If that can be done on sourceforge
(as it can, to some extent, see GNADE (though I am currently getting
operation timed out for the home page)), then so be it.
And if someone has the time to do it there, and if there is no
risk of loosing anything that you do not want to loose (rights,
the connection, ...), then so be it.


: Yes, but one must first be able to *find* your page. Mustn't one?

The whole point of good hyperlinking is that you won't have to
do a search, because the page is linked in context :-)

:> Now think about the transitive closure of good hyperlinks.
: 
: Let's see
: 
: Searching:

  [...]

: OK I give up, maybe you do not have a homepage? ;-)

Two, actually :-) But for now one of them might only be accessible from .de.
see www dot gelumps dot de, a tiny update to follow in some days
(Friday probably :)

: I really don't understand why registering your homepage at freshmeat of
: yahoo or adapower is careless?

I didn't mean this, I have failed at saying that a freshmeat
centered network is not the solution I would prefer to a solution
where people take care of linking their well-written contributions
to useful information. One emphasis is on taking care.


Look at how this works with component libraries in c.l.ada.
We see some of their authors discussing aspects of their
products. (Thanks for this, BTW!) On the other hand, lack of
documentation is mentioned every now and then. Now imagine a
situation where authors put their thoughts and arguments right
into web pages.  Arguments could actually contain links from pro
to con pages (perhaps not "deep down" in type descriptions, but in
introductory or prefatory notes). This might require cooperation,
and abiding by http://www.w3.org/Provider/Style/

Effect: Once you are hooked into the Ada components web, you can
easily browse one library's documentation, and also move from one
library to the other. (This internal and external linking is one
reason why I was referring to coherent and coupled software modules,
I think.)


-- Georg