Hi all,

I was wondering if you could help me brainstorm an idea I just had.  I was
reading Ron Jeffries' book, online at the http://www.xprogramming.com site.
So, reading his book, although the references are all currently screwed up,
I came up with the idea of an output wiki.

An output wiki would be like a regular wiki in just about every respect.
You type stuff in according to formatting rules, everyone gets to
participate, and stuff.  But, each page has an owner, and there are "root"
pages.  Root pages are like tables of contents, and whenever someone starts
an e-book, he adds (and owns) a root page.  He can choose other editors as
well.  So can the Wiki administrator.

The first thing the person does is write the part of the e-book he wants to
write.  He then publishes that bit, and adds it to the root page.  From then
on, it looks and acts like a regular Wiki.  The major difference is that the
root page has an option to format and output a book (in postscript, html
help, TeX, or whatever).

While the thing is being built, most wiki pages won't be added to the root
page.  And most comments won't become part of the book text.  The editors
get to choose what parts of the Wiki get elevated to the status of book
text.  They do this by adding links to the root, and by modifying the pages
themselves.

By convention, the comments stay on the Wiki, always.  Also, there is a
signature/reference format for people to sign their comments.  This way, as
the book develops its form, the people who contributed can be recognized.
So, if the *Do it in a workspace* page has a comment by Kent Beck, saying,
"I also like to write code in the debugger," when that comment gets elevated
Kent's attribution does too, and when the book is printed he's in the
reference section.  Also, those people who commented without their words
being elevated will show up in the acknowledgements section.

The outputter does some trickery based on the style of output.  For example,
if the thing outputs MIF format files, then they will have the hyperlinks
defined for screen viewing and they will have "(see page n)" references as
well.  If the thing outputs straight HTML, then it will only have the
hyperlinks.  If it outputs Postscript, it'll only have the "(see page n)"
references.  All of these references can be indexed.

Sometimes, a page shouldn't be elevated to root status, but it is a
definition or something.  In those cases, special formats can be created, in
the when-you-need-it style.

So, I'm looking for more brainstorming, in case anyone has some ideas or,
perhaps, in case people think I'm barking up the wrong tree.

-John