Thanks to VIBBER that showed me the right pages, what I called domains seem to already exists (or at least, seems to be already in progress) under the name categories. So no more need for this page.

Goals

edit

The goal of this page is to present a new concept named domain to be able to extract a sub part of a wikipedia as an automatic process.

This is a proposal for comments. If I'm the only one to find it usefull, there is no need to implement this !

Imagine you want to make a wikipedia on a handheld PC. Disk space is a critical ressource on those computers, and you know you (or the ones who will use this wikipedia) don't really want to use the whole wikipedia, but just a sub part of the wikipedia.

Let's say, you want to make availlable on a handheld only the page related to History because you know a lot of historian would be interested in that stuff.

There is currently (if I'm wrong, please, tell me how to do that) no way to find all the pages related to the domain History.

You can imagine getting all the pages from the History page with a recursion level of 2 or 3 or even 4. But whatever tricks you'll use, you'll still have a lot of page that aren't related to History, and you will miss a lot of ones that really are related to History.

Of course, you may want also to extract all mathematical pages on french wikipedia or all geographical pages on polish wikipedia as well.

What wikipedia currently lack is a way to associate a domain to a page.

Requirement

edit

Those points are few idea about what domain should be or shouldn't be achieve the goal:

  • There must be a way to associate a domain to a page.
  • A page should not be restricted just to one domain. this would be too rectrictive.
  • Creation of a new domain should be easy. Associating an existing page or a new one to a domain should be easy as well. Otherwise domain won't be used.

Reflexion

edit

This section is to discuss some points. If you don't like the proposal in the next section, here is some stuff to propose better solution.

What is a domain ?

edit
  • There must be a domain list somewhere. If someone want to add one or several domains to a page, he should be able to list all the domains to find the good ones, otherwise there will be too many useless domains.
  • There is no doubt about a domain should be a page. There is three main propositions about what a domain could be :
    • A domain can be simply a page like History where you selected a combobox this page is a domain in the edit page.
    • A domain can be a page in a namespace like Domains:History. All valid domains are all pages in the namespace. You can then imagine a page like Special:Domains where you'll find the list of all domains
    • A domain can just be a domain because at least one other page refer to it as it's domain. For exemple if World War II say that History is it's doamin (or one of it's domains), so History will be a domain.

How to associate a domain to a page ?

edit
  • Associations should be easy.
  • I can see two main ways to do it :
    • Adding a selection list on the page named Add a domain to this page.
    • Inseting a special link to a domain within the page would associate the page to the domain. That would be exactly like the interwiki links.

Proposal

edit

Here is my proposal. Feel free to give your opinion.

  • Domains are located in a namespace. (for example domains:)
  • A domains is valid once there is at least one page that link to the domain name within the domain namespace. (if World War II link to Domains:History, History is a valid domain)
  • The page Domains:History will contain (automatically with the "What links here" internal function) the list of all page within the domain.
  • Editing Domains:History can be disabled because there is no need to. Informations about History can be put in History while information about which page to put or not to put in the domain History can be written in Domains_talk:History, the talk page associated.
  • The list of all domains would be listed in something like Special:Domain_list
  • The list list of all pages without domain (to make it easy to update a wikipedia to use domains on all pages) may be accessible within a page like Special:pages_without_domains

Comments

edit

Thank you for reading the whole pages.

Comments are welcome on the discussion page Talk:Using_domains

This sounds rather like categories, code for which is being hashed out right now. See ongoing discussion of implementation details on the wikitech-l mailing list. Test code is running on http://test.wikipedia.org/ --Brion VIBBER 22:13 28 Jun 2003 (UTC)