Could also use the magic word NOINDEX
See http://www.mediawiki.org/wiki/Help:Magic_words

I don't think "unindexing" content is the best solution here.

First, disabling crawling doesn't clear the index. It take a long time for that information to expire.

Second, that content may still be useful to someone. That's especially true of the 3.X documentation.

Third, I'd like to see what the best practices are. For instance, I get old versions of the VTK docs all the time when I do a search. Same for python sometimes. That's not great, but it is predictable and understandable behavior.

The alternative fixes are to

1) provide pointers to the most current documents at the top of the 4.0 pages (that's what W3C standards do -- always maintain a "latest" pointer)

2) Have a "latest" version on the wiki that redirects or is copied in some way,

3) Have documents that describe current behavior as well as recent changes (for example, the python docs say "since version 2.6" or "added in 2.5").

Good point. Will think about it. Assigned the issue back to me.

From Lauren - May 3, 2012:

you could do what NIH does for old grant info pages, or what amazon does so that people know the model number is old, edit the top of the page with info pointing to the current version

or have wiki URLS that point to Current, and redo those links when the latest stable version changes, maybe they will accumulate more google cred since they are longer lasting URLs

my 2 cents

Closing the issue because the approach used in 0002887 achieves the same purpose.

Closing resolved issues that have not been updated in more than 3 months

Documentation references version numbering. See https://www.slicer.org/wiki/Documentation/4.6/HowTo

Documentation references version numbering. See https://www.slicer.org/wiki/Documentation/4.6/HowTo

Does this mean only 4.7 documentation is indexed now ?

No, I'm saying robots.txt does not inhibit indexing of documentation. The principles laid out in the HowTo describe the structure, versioning and navigation of documentation resources for all versions of Slicer.

Sorry for the noise. I initially thought you closed the issue ... but I did a while back. That is why I got confused ... oops.

That said, do you know of way to favor indexing of a given set of pages ?

Make sure it's mobile-friendly. And use meta tags to ensure that relevant search terms are present aka "Documentation" is on the right page set and "Old Documentation" is on the prior content. Other than that, I suppose you could use Webmaster tools to see when and how much Google is indexing the site.

When I search for "Slicer documentation" I get https://www.slicer.org/wiki/Documentation as the 0000001 result - so I guess we're doing something right.

I looked at how readthedocs was dealing with this and they make use of canonical urls

http://docs.readthedocs.io/en/latest/canonical.html
https://moz.com/blog/how-to-advanced-relcanonical-http-headers
https://en.wikipedia.org/wiki/Canonical_link_element

For any given page https://www.slicer.org/wiki/Documentation/X.Y/subpage[/subpage[...]] If we could automatically include the following either

• as link directly into the html

<link rel="canonical" href="https://www.slicer.org/wiki/Documentation/Latest/subpage[/subpage[...]]" />

• or as a response header.

I think we should be all set

Also to implement the Latest two approaches:

1. rewrite url on the web server
2. explicit page creation on the wiki

I think (1) would be easier, this also mean that at each release of Slicer we would have to make sure to update the rewrite rules.

Thinking more about it, updating Latest pages ensuring they redirect as expected directly on the wiki is also feasible.