homepage

The Python version selection for the documentation should be cached.
It's very annoying having to select the preferred version each time one follows a link, e.g. search result, irc post, etc.
I'd like to see caching the preferred version in a cookie and automatically switching to this python version when opening doc pages.

I don't think that makes much sense. People often link to a particular version of the documentation and making the version "sticky" to a particular user could be confusing.

it should rather be a opt-in feature. and when the redirection triggered, one should be able do click (you know it from wikipedia) back to page where one was redireced from.

On Sun, Aug 24, 2014, at 16:39, Jonas Jelten wrote:
> 
> Jonas Jelten added the comment:
> 
> it should rather be a opt-in feature. and when the redirection triggered,
> one should be able do click (you know it from wikipedia) back to page
> where one was redireced from.
That would interact poorly with linking to particular sections with a #.

I agree that this is a bad idea.
1. Cookies are generally a nuisance. I like the fact that docs.python.org is (apparently) cookie free and that python sites only use a few short-lived cookies. The complications you propose would be a nuisance to create and maintain.
2. Not too long ago, we created docs.python.org/2/ and /3/ so people can use generic links to the latest docs for the latest released version of either Python 2 or 3. The aim was to reduce the problem of links to stale docs. Both sets of docs have version added notes for changes within a series. Those two docs are not interchangeable and should not be automatically switched.
If someone links now to a fixed version of the docs, they are either ignorant of the new system *or* they have a purpose that should not be over-riden.
If someone quotes or paraphrases a portion of a page (which is a claim about what the page says) or otherwise makes a claim about Python, and posts a link as evidence of the claim, clicking the link should take one to the 'evidence', not one's preferred version.