specifications

["Greg O'Keefe" <gcokeefe@postoffice.utas.edu.au>]

Hi,

Someone asked recently for dreams about the LDP. Here's a thought.

Open source software is often produced by hacking rather than by
a disciplined procedure. There are often no specifications for
the software we use. Specs are good because they set goals for
software, against which its performance can be measured. Also
good specs could help people decide what sofware can best satisfy
their needs. 

If specifications were available in a clever format, perhaps some
for of sgml (uml?) some documentation could be generated from it.
As could tests. Specifications could be made for existing software,
by reverse engineering, or they could originate from the ldp.

This may be going beyond the usually conceived aim of the ldp. 
But must documentation always be done after the fact? Why not
initiate a project with specifications? Perhaps the ldp could
gather information about people's needs and produce specs for
programs that would fulfil those needs. 

The main problem I see, is that free software authors just don't
want to work that way. They code what they feel needs coding,
not what someone else told them to code. But maybe there are
enough developers who can work to specs to get some projects 
going this way. And if they are sucessful, the idea might 
catch on. 

Sorry I can't be more concrete. The basic idea is that the ldp
become a generator of ideas for free software rather than
reacting to what exists by explaining how to use it.

Greg O'Keefe


--  
To UNSUBSCRIBE, email to ldp-discuss-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org