Talk:Community Wishlist/W356

Thank you for your proposal

Hello @Stjn, thank you for this proposal, I will let you know if we have anything we want you to clarify and also if it does progress into a focus area. Have a good week. –– STei (WMF) (talk) 20:25, 29 January 2025 (UTC)

Follow-up: Clarification Needed

Hello @Stjn,

Thanks for the suggestion. It’s a good idea, but some parts of the request are unclear or conflict with how CirrusSearch works. For example, adding a keyword to exclude redirects might be simple, but changing how redirects are indexed or searched across all queries would be more complex. We recommend clarifying the request and specifying which behaviors you’d like to see—this will help us assess it more accurately.

--ARamadan-WMF (talk) 12:50, 3 April 2025 (UTC)

@ARamadan-WMF, respectfully I am not sure I understand the comment at all. The request is pretty simple: currently it is impossible to either include or exclude redirect pages in searches, they are inconsistently either included without a possibility to exclude them (which is bad when you want to look up title matches only in actual articles, for example) or without a possibility to include them (which is bad when you want to search for templates used exclusively in redirects or for presence of some things in their source code). What exactly needs to be clarified here? stjn^[ru] 20:23, 3 April 2025 (UTC)

Hello @Stjn,

Thanks again for your follow-up and for clarifying the intent behind the request.

We discussed your proposal with the responsible team, and their current assessment is that making this change would introduce significant complexity to how Search is implemented. Since redirects aren’t indexed directly in the current system, enabling this feature would likely require several months of development work.

That said, they’re open to investigating it further if there's a strong enough use case or demonstrated need from the community.

Let me know if you’d like to expand the proposal further in light of this. ARamadan-WMF (talk) 12:30, 16 April 2025 (UTC)

The need was already demonstrated ad nauseam in the Phabricator task. The current way it works makes it impossible to do routine searches properly (such as intitle without redirects or insource with redirects). I am not sure what sort of broken telephone is at play here, but the fact that it would supposedly take a lot of work to include redirects in the search engine (which is a bit confusing to hear, considering indexing things properly shouldn’t require significant complexity in the current system) doesn’t necessarily mean that there isn’t any demonstrated need for it. I guess I can at least say definitively that the new Wishlist process does not necessarily help to solve problems any more than shouting into the void on Phabricator does. Thanks. stjn^[ru] 16:57, 12 May 2025 (UTC)

@ARamadan-WMF, Could you explain how this will (at a technical level) add significant complexity to search? (Any phabricator tasks or mediawiki pages would be helpful). Sohom (talk) 16:45, 2 December 2025 (UTC)

The short answer is that the data model for search does not consider redirects to be their own page. Redirects, within the search index, are a property of the page they redirect to. Changing the fundamental structure of the data model is not something that can be done in a couple weeks by an engineer or two, it is a reimagining of how data is stored in search. We could potentially add a modifier or variant of intitle that skips the redirects, but the current structure of the search index does not allow for treating redirects as independant things across all of search. Ebernhardson (talk) 19:08, 5 January 2026 (UTC)