[SEO] Make the homepage's canonical link tag always point to the homepage URL

[sutibun]

1. Installed Backdrop.

2. Created a Page with URL "/mysweethome" and set that as the frontpage.

3. When I looked at the source code for the homepage, I noticed the canonical link tag is pointing to the Page's URL as follows

   <link rel="canonical" href="http://www.example.com/mysweethome" />

This is wrong because search engines will now think the homepage is /mysweethome. Not good for SEO.

Regardless of the Page, the View, etc, the canonical link tag for the homepage/frontpage should always use the homepage's URL.

<link rel="canonical" href="http://www.example.com/" />

[quicksketch]

Thanks @sutibun! Very correct. The accidental revealing of the home page "path" drives me crazy. The homepage should never have a path at all.

This issue also mildly relates to our core "Global Redirect" issue at #905. If implemented, that issue would also prevent accidental visiting to the homepage path, and redirect you to the canonical URL.

[sutibun]

@quicksketch Already having to deal with other SEO issues from switching to Backdrop, I was so worried last night about that wrong canonical link tag that I tried everything to remove it. Especially since didn't know how to customize it.

Long story short, I stumbled upon a way to do so. Maybe as a tip for others and as a stop gap, simply create a unique Layout for the homepage and use blocks for content. While there is no canonical link tag on the homepage using this way, that is much preferable to one that is pointing to the wrong location.

[docwilmot]

Aside, we really need a place for tips and tricks type stuff like this. Can't go on the API site. I had already offered to start building user docs, willing to continue if we can get some sort of action plan with a timeline commitment. I had suggested we can build the docs wiki style and transplant them once they're ready. Started on my google drive site but neglected till @quicksketch or @jenlampton is ready to roll.

Sorry to hijack. Carry on.

[sutibun]

Just point the monkey where to add stuff and will do so.

[quicksketch]

While there is no canonical link tag on the homepage using this way, that is much preferable to one that is pointing at the wrong location.

This is an interesting question just in general. Should the homepage always have a canonical tag? I would say yes, as you can access a path at http://example.com, http://example.com/index.php, http://example.com/index.php?q=node, http://example.com/?q=node.

But right now we only add canonical tags on node pages, which seems surprising considering any path in Backdrop would demonstrate these same problems.

[sutibun]

If the CMS isn't going to implement canonical link tags well, IMO, it's better there is none. Google can figure out which is the canonical URL without those link tags. Remember, canonical link tags haven't always existed and the web did fine. (i.e. they're nice to have and can help ambiguous situations but not necessary)

By introducing incorrect canonical tags, you start misleading and confusing Google. I rather not play that game... I think of canonical tags as something like 301 redirects.

[quicksketch]

But right now we only add canonical tags on node pages, which seems surprising considering any path in Backdrop would demonstrate these same problems.

Hm, the reason for this is actually fairly clear. Because a lot of pages have pagers (including the default homepage of /node), the query string modifies the content of the page. e.g. /node?page=2 may have very different content than /node. So you don't want /node to be considered canonical for the subsequent pages of results.

By introducing incorrect canonical tags, you start misleading and confusing Google.

Yes, my thought exactly, let's hold off on trying to make some global solution if it's only going to cause a problem.

[quicksketch]

PR filed at backdrop/backdrop#1103 that solves the homepage problem.

This is great because it incidentally fixes calls to url('node') (or whatever the homepage path is). Now if it is the homepage URL, it will always be trimmed down to hide the internal path.

[sutibun]

sweet. ^_^)b

Yay, I finally made an indirect contribution

[klonos]

If implemented, that issue would also prevent accidental visiting to the homepage path, and redirect you to the canonical URL.

I guess that /homepage/edit would still have to work. As any other /homepage/* path. Right?

[Dinornis]

Looks like this issue needs some attention. The problem still exists in 1.5 although looks like @quicksketch solved it in PR 1103 which I can't access to test it.