Subject:
Re: Wikipedia articles on the LDP
From:
David Lawyer ####@####.####
Date:
28 Jan 2002 11:00:17 -0000
Message-Id: <20020128030050.A711@lafn.org>
On Sun, Jan 27, 2002 at 03:31:30AM -0500, David Merrill wrote:
> On Thu, Jan 24, 2002 at 10:10:34PM -0500, David Merrill wrote:
> > I want to start mirroring some relevant Wikipedia articles on the LDP.
> > While the LDP has a pretty good (and growing!) collection of HOWTOs
> > and Guides about the technical issues, we have very little content
> > about the larger concepts. These articles would help fill in the gaps.
> > And as a nice bonus, they are already all well maintained and getting
> > better every day. We have Wikipedia's blessing to use them.
>
> It's been three days, and no response. I don't know how to interpret
> that. Does nobody care, or does silence mean agreement?
>
> The organic nature of the LDP as an organization means I try always to
> get consensus from the major contributors before I do anything major
> that would affect the site. So I really want to hear from you. Please.
>
> Here's the gameplan I'm pursuing. I hope to have lots of new authors
> come on board who will want to use the WikiText editing method,
> because it is so very accessible and easy. But it also has some real
> positive advantages, as well as some capabilities, that DocBook alone
> doesn't have.
I just don't like the concept of wikis. However, using the wiki markup
for writing howtos is another question. I read your Wiki-HOWTO and I
think that linuxdoc-sgml is just as easy. It's very important to make
it easy on authors so I agree with you on that point. One thing wrong
with the wiki markup is that it's not at all like docbook. Linuxdoc
markup is something like docbook, only simpler. So if someone wants to
eventually learn docbook, starting with linuxdoc will help more than
starting with wiki-markup.
Wiki markup reminds me of sdf (simple documentation format) which has
many more "tags".
----------------------------------------------------------------------
One idea for writing certain short articles in plain text (with no
markup at all) is the following: Someone writes a short article which
tells how they did something (how they configured something, fixed a
problem, etc.). They email it to the author of a HOWTO that covers
their topic. Then if the author approves, a copy is sent to LDP in
plain text (or html) and put on the LDP site, but not made a part of the
HOWTO. The HOWTO author creates a link to the article from the HOWTO.
This would be used only for articles that would be of interest to say
only one person in a thousand that reads the HOWTO. If it's of general
interest, it should be added to the HOWTO (after editing by the HOWTO
author --possibly after being completely rewritten). If the article
gets to many hits, then the HOWTO author would be contacted so as to
incorporate the article into the HOWTO.
Something similar to this happens now when someone emails a HOWTO author
and sends the author the url of a website where such an article is.
Then the author puts a link to it in his/her howto. The problems with
this are: 1. Some people don't have websites 2. Personal websites often
are changed or go dead.
This suggestion is not something that is urgent. I think there are
other projects more important such as making linuxdoc (or a subset of
it) convert cleanly into DocBook. Then we could suggest LinuxDoc in the
LDP Author Guide (at the very start of it so that they don't need to
read anything about DocBook).
>
> The namespaces concept is one of them. I let you make links using text
> like this:
>
> [[wiki:Linux kernel]]
> [[ldp:Linux-FAQ]]
>
> and my code resolves those to be links to the correct urls. Now what
> I'm going to do with the Wikipedia articles is first select a set of
> what I consider to be good articles for the LDP, and then do an
> automatic update that looks at all of our WikiText documents, sees
> which Wikipedia articles they link to, and mirrors them. All nice and
> automagic, the way I like things.
If there are a lot of links, and you follow the links recursively then
you may wind up with a very large number of files, many of which could
be of poor quality.
>
> This code is all written.
If you already have the code, it makes it hard to object. But I've got
a major concern. In any organization like the LDP, there should be at
least 2 people that fully understand any particular function. If David
Merrill suddenly leaves LDP (and such situations are frequent, only a
few years ago the LDP leader suddenly quit) then who will understand his
programs and continue maintaining them?
For example, txt2db doesn't covert text to docbook, it converts a wiki
format to docbook. I have a bias in favor of standard tools provided by
major distributions such as Debian. Well, perhaps txt2db could (after
being renamed wiki2db :-) become part of a Debian package.
> I'm extracting the links, collating them from the Wikipedia data, and
> converting them from Wikipedia's format to my WikiText. Their format
> is similar, since it's what I based mine on, but it is not identical.
> Among other things, they let you put a sect3 without a parent sect2
> tag (since it's only using html headings). So there's a translation
> step, which is only a few lines of perl.
>
> Now I'm going to write a script to generate an index.html, create a
> /wikipedia/ subdirectory, and add a link on the home page. If you have
> problems with this, or concerns, or just want to talk about it first,
> then speak now or forever hold your peace. :-)
>
This is off topic, but related. The xml LDP database has all the
defects of docbook :-). I suppose one may claim that since it's not
edited directly, that it doesn't matter. But even if it's not edited
directly, it could be a lot shorter if it used short tags (and no end
tags) like linuxdoc-sgml. There are a number of other databases but I'm
not familiar with them. Would any of them be better (assuming that we
were not already into xml for our db)?
David Lawyer
