Every Rutba page can tell you which page it is

A support conversation about a website usually begins with the same three questions, and the person reporting the problem can rarely answer any of them. Which page, exactly? On which of our sites? And when — before or after last night’s release?

"The login page is broken" is a real report, sent in good faith, that describes one of several pages on one of several deployments of the same code — several of which look identical on purpose. Answering it starts with a round of questions nobody enjoys.

A reference on every page

Every page on every Rutba site — and in every Rutba app — now carries a short line like this:

auth.rutba.io  ·  AUTH-LOGIN-PASSWORD  ·  0.1.0+a3f19c2  ·  d4dd01b8
 which site          which page          which build     which visit

• Which site — the deployment that served the page, which matters because several of them run the same code and look alike on purpose.
• Which page — a stable, readable name. It never changes once a page ships, because it is something people write down.
• Which build — what tells "it worked yesterday" apart from "it is broken since the release".
• Which visit — the same request id our logs and error messages carry, so a screenshot is enough to find the exact request.

Tap it and the whole reference selects at once, rather than one word of it, because the point is that you paste it into a message.

A screenshot of the footer is enough. Nobody has to answer "which page, and when?" again.

A reference must never carry your data

This is a line people are actively asked to copy into support tickets, which makes one rule absolute: it must never contain anything private. Many addresses include a value that belongs to somebody — a quote’s private link, a one-time token.

So the page name is built from the address with anything that looks like a value dropped: numbers, long codes, anything that reads like a token. The rule leans one way on purpose. Dropping a segment that was harmless costs a slightly vaguer reference; keeping one that was a token would leak it into somebody’s inbox.

It names a page, not a record

Every post on this blog is the same page with different words in it, so they all share one reference: a blog post page. A reference per post would be one nobody could quote or search for. A post page still reads differently from the blog’s index, so "it is broken on the blog" never names two pages at once.

Rendered above everything, so no page is missed

The first version of this was on the pages somebody remembered to add it to, which is never all of them. It is now rendered from the one place every page on a site passes through, so a page cannot opt out by accident — and a page whose address cannot describe it names itself instead.