xml structures and some notes

Thomas 'Balu' Walter tw at itreff.de
Thu Sep 7 07:40:06 PDT 2000

+-Gerard Beekmans-(gerard at linuxfromscratch.org)-[07.09.00 05:01]:
> > And you start doing that now, and we will never start coding (i'm serious)
> > there's a reason you test something to see if it works, The frist time
> > around it doesn't have to be perfect, it just has to work, imo
> Like the lfs book evolved. From one big sgml file to mutiple files, each 
> chapter in it's own files, to every section it's own file. The book got split 
> up from 1 file to it's current form that contains aprox. 425 sgml files. Why 
> did it take so long: Same what Jesse told: you need to have something that 
> works. 

It would not have taken so long if you thought of it before you started
the book - perhaps you would have noticed that "the book will get bigger
and small files are easier to handle" and start directly with those
smaller files...

> weeks I'd like to start coding. Sure, in a year the tools will be totally 
> different but that's ok. It's called evolution. It takes time.

You call this "software-engeneering" and as the book "Web
Application Development with PHP" says: 

It's not necessary to inform the user of the underlying system, wether
you're using Julian or Gregorian dates or maybe even your own format -
at some point, you might want to provide an exta set of such features to
the user (for example date format conversion), but it's completely
unecessary when all you need initially is simple to enable someone to
manage appointments.
On the other hand this doesn't mean preventing or even disabling a
future implementation fo these features. The trick when designing an API
is to meet your requirements exactly, while being able to extend the API
to any eventual needed functionality. This requires in-depth planning
and thoughtful definitions, as discussed throughout this chapter.


More information about the alfs-discuss mailing list