launchpad-dev team mailing list archive

[Martin Pool <mbp@xxxxxxxxxxxxx>]

I was a bit inspired by the recent Google developer day to do
something about the poor results that some times come up when you
search for pages in Launchpad.

There are a few specific fixes and I also added a few new facilities I
hope you will think about using when you touch other pages:

1- canonical_url_recommended property on a view should be used when
there are several pages at different URLs that have essentially the
same content. "Essentially the same" is a bit subjective: it doesn't
have to be byte-for-byte, but the real test is that a user would not
want to get two or more pages that are essentially the same in their
search results.  So the same bug in different contexts is essentially
the same; the same list of things with different sort orders is also
essentially the same.

2- page_description can also be on a view (analogous to the title) and
should return a "the most important text" on the page, if there is
bulk text, suitable for putting in the description shown under the
search result.  For instance for a bug this is the bug description.
If you have nothing to say, say nothing: returning hardcoded text like
say "A list of branches" is probably bad.  This shouldn't duplicate
the title.  <https://code.launchpad.net/~mbp/launchpad/meta-description/+merge/82769>

3- page_created, page_last_modified - (not added yet but I think I
will) - if the page provides these, they will be mapped into
schema.org metadata.

4- schema.org microformats - these need to be done in the page
templates themselves; basically you give the page a bit of lightweight
semantic markup to say "this is a date" or "this is a comment by
person X on date Y."  Google increasingly parses these out and uses
them to generate better results and better snippets on the results.
<https://code.launchpad.net/~mbp/launchpad/888353-microformats/+merge/82767>

To some extent we're guessing how the black box of search engines will
behave - in the first cut I'm trying to do something large enough we
may see some difference without spending too much time or making too
intrusive changes.  All these are open standards.

If they do help I hope you will think about using them on pages you
touch too, so that searches across Launchpad are more useful to our
users and so Launchpad's a better resource for the internet at large.
What I am hoping for is:

 * Launchpad pages being higher ranked (when they're a good answer to
the question)
 * Better snippets on the pages that do come up
 * Avoiding unwanted dupes in results
 * More 'rich snippets' coming up on lp pages

-- 
Martin