Re: [DOCS] Canonical url for latest document set?

On Wed, Jul 12, 2017 at 11:04 PM, Conor McNally <theconor@gmail.com> wrote:

Hi,

There was a suggestion a few years ago on this forum about adding a canonical url to the documentation set:

https://www.postgresql.org/message-id/04E4F5A6-6526-4DDC-A9E5-2991E3B2ED83%40cantoute.com

Was there a reason for not doing this?  At the moment it can be pretty annoying googling something like "postgres select syntax", and getting a pretty random selection of duplicate pages, none of which matches the current version of Postgres.  Even google for "postgres 9.6 select syntax" still doesn't return the correct version for me.

Yes, and the reason being that the docs aren't actually duplicates.

Something like "postgres 9.6 select syntax" *should* work. But more importantly, if we make it the canonical URL, then it will be impossible to search for old versions at all, will it not?

 

In contrast, other documentation sites such as MySql, CouchDb, Python, Django, etc. all behave much better in search engine results.  For CouchDb and Django you can see use of the canonical tag on the documentation pages.  For the others they must be doing something different, but still they've managed to direct Google to the single, current version of the correct page.

Yeah, I'm a frequent user of the django docs at least, and it's absolutely horrible. Since I'm not using the very latest version, then *every* time I search for docs I get sent to the wrong version. So I've got basically 100% failure rate with the django docs, vs maybe a 50-60% failure rate with the postgres docs.

 

I'm just wondering if something has looked into this already, or if canonical url is something that could be tested for the postgres documentation?

The problem with testing it is that I assume its' destructive testing. Once we've tested it, it's very hard to go back.

 

Here is some information from google:
https://webmasters.googleblog.com/2009/02/specify-your-canonical.html

The goal for me would be to make the documentation for the current version of Postgres rise to the top of search engine results, and to remove the random results for older versions.

What do you think?

If we can find an actual statement from somebody who *knows* (and nto just speculates like most of us do) how it works.. In particular, will a search for older versions still work or will that force *everybody* to the current version?

Things do seem to get worse by the day. I guess we're eventually approaching the point where all version specific queries *already* fail. We're not there yet, but we seem to be closer than we've ever been before..

--

 Magnus Hagander
 Me: https://www.hagander.net/
 Work: https://www.redpill-linpro.com/