1457959 - Consider en-* pages always completely localized on production

Status: RESOLVED FIXED

(www.mozilla.org :: Bedrock, enhancement)

Description

[Francesco Lodolo [:flod]]

We're adding a new English locale to Bedrock (en-CA). With this, we have 4 English variants: US, CA, ZA, GB.

Right now, these locales are treated like normal locales, meaning:
* If a page is not activated, it will fall back to en-US URLs.
* If a page is not completely localized and missing tags, it will fall back to older content.

While we should give them a chance to change translations, I also think that we should treat them differently from other locales. For example, en-GB for 60 won't have a What's new page, since it's not localized.

My request would be to treat all en-* locales in PROD like we do on DEV: pages are always activated and tags available, even if the translation is incomplete. This way, we would keep the original URL, and only fall back at content level to American English.

Comment 1

[Paul [:pmac] McLanahan]

The problem I can see with this approach is that we would have more URLs with identical content and that would be bad for our search engine placement for those pages. I think :raue and :agibson were working on this at one point. I'll NI them.

Comment 2

[Francesco Lodolo [:flod]]

(In reply to Paul [:pmac] McLanahan ⏰ET needinfo? me from comment #1)
> The problem I can see with this approach is that we would have more URLs
> with identical content and that would be bad for our search engine placement
> for those pages. 

I don't think that would be worse than the current situation?

Comment 3

[Alex Gibson [:agibson]]

(In reply to Paul [:pmac] McLanahan ⏰ET needinfo? me from comment #1)
> The problem I can see with this approach is that we would have more URLs
> with identical content and that would be bad for our search engine placement
> for those pages. I think :raue and :agibson were working on this at one
> point. I'll NI them.

We resolved the duplicate content issues for en-GB by making some subtle (non-user facing) differences to the content. See https://github.com/mozilla/bedrock/pull/5461 

We may likely want to follow a similar approach here, but I'll defer to :raue

Comment 4

[Raphael Raue [:raue] [:rraue]]

It will cause more issues then we already have. The more duplicate content the more confusion for an search index. But I dont see a solution here, as we want to have these different landing pages. In an ideal world, we would have for every language one page and this page would serve the right download based on accept language and geo-location and if thats the wrong one, lets a user choose on the /all/ page. 

The problem is that Google is not clear on such pages and even change the apperance of such setups in the search results. The standard Google says is: 

- If the pages are completly the same, canonical should point to the main page, hreflang points to all language/country versions.
- If something is different like currencies the pages should be self canonicalized and hreflang points to all language/country versions

But both versions did not work for us and there is no third. Because what Google really wants is unique pages for everything, what would be just too much for just serving the download.

Comment 5

[GitHub Bugzilla PR Linker]

Attachment: Link to GitHub pull-request: https://github.com/mozilla/bedrock/pull/5613

Comment 6

[[github robot]]

Commit pushed to master at https://github.com/mozilla/bedrock

https://github.com/mozilla/bedrock/commit/5bdc83a841d4c3cc7d5c433cddd7f69d97771fac
Bug 1457959: Fix redirect tests for new en-* locales

Comment 7

[Axel Hecht [:Pike]]

If I'm reading this right, we currently disambiguate en-US and en-GB, but en-ZA and now en-CA both dupe-content-conflict with en-US as soon as their pages are up-to-date?

Comment 8

[Francesco Lodolo [:flod]]

(In reply to Axel Hecht [:Pike] from comment #7)
> If I'm reading this right, we currently disambiguate en-US and en-GB, but
> en-ZA and now en-CA both dupe-content-conflict with en-US as soon as their
> pages are up-to-date?

en-CA was covered in
https://github.com/mozilla/bedrock/commit/87726161d7e0921cf0f1555815024d8f3606023a

The question about en-ZA stands (even if we don't have an active team there).

Comment 9

[Francesco Lodolo [:flod]]

Assuming this was fixed.