[REBOL] Re: Rebol programs
From: bobr:resume:dprc at: 21-Mar-2002 11:09
Charles wrote:
> I missed the original post somewhere, so I'm quoting from Greg's post.
for those who missed it,
the original post follows my reply.
I have written quite a few heavy programs. I do not count lines.
I call each an application to differentiate it from a utility.
rebol as my solution foundation
Sometimes the application is actually several rebol scripts
which cooperate but run at different times or only when certain events
happen.
For example a rebol based CGI runs when the page is fetched,
another when the form completes and another later when cron must
do hourly processing and still another when distribution
to mirror sites is complete.
I could, in several cases, have built one monolithic script
which is re-called for each of these events but that is not my style.
If refactoring yields some common code then I can use 'do or 'include
to advantage. I add to that with 'default so that I can unit-test.
I simultaneously build up from the bottom (with small unit tested utils)
and down from the top for these major applications.
'include and 'default are add-ons refined with help from others here.
Applications built this way that I am allowed to talk about:
a router reprogrammer,
several newsfeed handlers
and a website 'enlivener'.
The first one goes out and collects data about equipment connected to
a certain model of router used for remote support. It reprograms the
router
if the support team needs to add equipment and make it NAT reachable.
Its an embedded application which cannot make use of rebol/view.
It knows how to self-update if the home site publishes revisions.
The enlivener application runs for every page on a website
and replaces sample values with live data much like xml promises to do.
It has the advantage of not requiring a format change to xml when
upper management at the bank approves a page layout alteration.
It replaces perl and mod-perl for performance with a lower cost
of maintenance.
The web design company does not have to send me xml,
they upload static pages as templates right to QA.
QA sees a live page with real data.
If QA approves the change, the same engine
is also used in production as well.
In this final phase the names of web sites in URL references change
to get images from some image server (like akamai) based on
rules the bank laid out months ago. Same with nav links
that would normally point to the design company server or
to QA servers.
I did not do the whole site with rebol functioning as the server itself
because it had to cooperate with other legacy apps already there.
I did not have the luxury of rebol/command at the time.
[riusa--email--it]
wrote:
