Page MenuHomePhabricator

Make sure removal of the .m domain work as it is expected for it.wikipedia.org
Closed, ResolvedPublic3 Estimated Story Points

Assigned To
Authored By
Peter
Sep 4 2025, 5:06 AM
Referenced Files
F66261667: Screenshot 2025-09-23 at 13.52.20.png
Tue, Sep 23, 11:54 AM
F66261237: Screenshot 2025-09-23 at 13.50.45.png
Tue, Sep 23, 11:54 AM
F66095658: it.wikpedia.org-direct-trace.json.gz
Sep 22 2025, 12:50 PM
F66095657: it.wikpedia.org-direct.har
Sep 22 2025, 12:50 PM
F66095584: search-it.wikipedia.org-after.har
Sep 22 2025, 12:50 PM
F66095582: search-it.wikipedia.org-after..json.gz
Sep 22 2025, 12:50 PM
F65957997: it.m.wikipedia.org-trace-1.json.gz
Sep 4 2025, 8:08 AM
F65957996: it.m.wikipedia.org.har
Sep 4 2025, 8:08 AM

Description

Let's collect real traces from a phone before we do the change and after we did the change. I'm thinking of two different user scenarios:

  • direct access to a wikipedia article using the desktop domain, using a mobile phone
  • click on a wikipedia link on Google search result, using a mobile phone.

Lets do it on https://it.wikipedia.org (the change happen 8-11 September)

AC:

  • Collect a HAR and trace for the two use cases for it.wikipedia.org before the change and attach them to this task
  • Collect a HAR and trace for the two use cases for it.wikipedia.org after the change and attach them to this task

Event Timeline

Peter set the point value for this task to 3.Sep 4 2025, 5:07 AM

Attaching a HAR and a Chrome trace (drag and drop into the performance tab) collected using a Samsung A51 for the search scenario.


And direct access, through the "old" desktop domain:


And using the m.-domain:


Peter renamed this task from Collect HAR/trace before and after removal of the m. domain for it.wikipedia.org to Make sure removal of the .m domain work as it is expected for it.wikipedia.org.Sep 4 2025, 1:40 PM

@Peter FYI: Unified mobile routing was enabled for it.wikipedia.org on Wed 17 Sep 2025 (T403510).

Thank you @Krinkle ! All looks good I'll write a summary tomorrow and close the task.

Search using "old" desktop domain


Access using "old" desktop domain


This works as it should.

Here's some screenshot from the waterfall where we automate going to Google, do a search and click on the result, going to Wikipedia. Before we did the change coming from Google the browser did these requests. You can see the redirect there:

Screenshot 2025-09-23 at 13.50.45.png (230×1 px, 60 KB)

With the fix, the redirect is gone:

Screenshot 2025-09-23 at 13.52.20.png (184×1 px, 48 KB)