[hibernate-dev] Hibernate site SEO optimization

[Sanne Grinovero]

On 10 December 2015 at 15:26, Steve Ebersole <steve at hibernate.org> wrote:
> I am not really sure what "have their headers patched to declare that the
> new guides are the canonical version" means in practical terms...

It means that the HTML pages of the documentation should have some
metadata within the header section, to point to an URL of the
"canonical" page.

This is interpreted by indexing robots like Google to let them know
which one single URL should be considered the "best" place to point
users too; it's also important as it means that we acknowledge that
the fact that they contain similar content is intentional: failing to
do so will have you suspected of plagiarizing good content from other
writers, and get your score within the search engine penalized, or
even fully banned from search results.

The technical details are described here:
 https://support.google.com/webmasters/answer/139066?hl=en&ref_topic=4617741

So that implies we should patch the existing uploaded documentation
pages, except all those old enough to warrant being completely
delisted from the root. If we can't access the root sitemap on
docs.jboss.org, then we should patch them all.. hopefully with some
sed and rsync shouldn't be too much work to run it on all of them.

Sanne

>
> On Wed, Dec 9, 2015 at 10:55 AM Sanne Grinovero <sanne at hibernate.org> wrote:
>>
>> On 9 December 2015 at 16:36, Steve Ebersole <steve at hibernate.org> wrote:
>> > We could leave the ones that are on docs.jboss.org in place and just
>> > move
>> > the current stable ones.  I see no benefit to moving ORM 3.2 docs over
>> > e.g.
>>
>> I'm ok to leave them there, but if we do they at least need to have
>> their headers patched to declare that the new guides are the canonical
>> version.
>> Otherwise they might be flagged "duplicated content" as I guess many
>> paragraphs didn't change much over time (or at all), and since the
>> position on docs.jboss.org is much older than the copy on
>> hibernate.org, that would penalize the score of hibernate.org
>>
>> >
>> > On Wed, Dec 9, 2015 at 10:08 AM Sanne Grinovero <sanne at hibernate.org>
>> > wrote:
>> >>
>> >> On 9 December 2015 at 13:54, Hardy Ferentschik <hardy at hibernate.org>
>> >> wrote:
>> >> > Hi,
>> >> >
>> >> > On Wed, Dec 09, 2015 at 03:12:54PM +0200, Vlad Mihalcea wrote:
>> >> >> Is there any change we host the docs on hibernate.org and simply
>> >> >> redirect
>> >> >> the old pages from
>> >> >>
>> >> >> http://docs.jboss.org/hibernate
>> >> >>
>> >> >> to
>> >> >>
>> >> >> http://hibernate.org/orm/documentation
>> >> >
>> >> > That would be an option I guess. Host the documentation somewhere
>> >> > else.
>> >> > I am sure
>> >> > we discussed this before, but I cannot remember the outcome.
>> >>
>> >> Right, this isn't new but for some reason (probably time?) we never got
>> >> to
>> >> it.
>> >> I thought the outcome was that we couldn't move the docs away from the
>> >> jboss.org servers, but Emmanuel just told me there's no such rule.
>> >>
>> >> So let's plan to move them all!
>> >>
>> >> I think we need:
>> >>  - redirects in place from the previous location
>> >>  - make/update/verify a sitemap for hibernate.org
>> >>  - get the docs on hibernate.org
>> >>  - insert the right metadata in the older docs so to point to the
>> >> latest as their "canonical form"
>> >>
>> >> And like Vlad suggested, let's simply exclude from indexing all very
>> >> old guides and docs which aren't easily patched to have the right
>> >> pointers to their latest version (I'm thinking especially about old
>> >> guides which don't have a 1:1 match).
>> >>
>> >> I could spend a bit of time on this during Christmas, not before.
>> >>
>> >> Sanne
>> >>
>> >> >
>> >> > Regarding hibernate.org. The website is build via Awestruct from this
>> >> > repo
>> >> > https://github.com/hibernate/hibernate.org and pushed to GitHub Pages
>> >> > for hosting.
>> >> > The simplest solution would be to add the documentation to the site,
>> >> > but
>> >> > that might blow up the repository too much and make syncs to
>> >> > production
>> >> > too slow.
>> >> >
>> >> > I am not familar whether one could upload the documentation to GitHub
>> >> > Pages
>> >> > into a separate directory. I am not familiar on what and what not can
>> >> > be
>> >> > done
>> >> > here.
>> >> >
>> >> > Either way, the upload of new documentation needs to be scriptable,
>> >> > since pretty
>> >> > much all Hibernate projects have a script of some form to upload the
>> >> > documentation
>> >> > of a release to documentation server.
>> >> >
>> >> > --Hardy
>> >> >