specifically, for "incategory" and "deepcategory" (i.e., pages _not in_ category X)
also, would be nice to support "hastemplate"
often times, you want to find pages in category X which do not include template Y and vice versa.
as a side: we tried to build the "poor man's version" of this feature in hewiki years ago, and it didn't take. this implementation is way more usable, and i believe many users will find it very helpful. kudos!
peace.
Good points. For the Wikisources where we more broadly utilise namespaces for content, and prescribe various templates for each namespace, this would be most useful.
For instance I was looking to identify pages in the Author: namespace that did not have the s:en:Template:Author, or not categorised into s:en:Category:Authors by alphabetical order. One would think that would be a reasonably easy query, I didn't find it so. [Yes, search is often more designed for reading and finding, not maintenance, that I understand.]
Hi @Billinghurst, thanks for the great examples! Supporting this sounds useful, and generally we want to look more into it at some point. Currently we focus on fixing bugs etc. though, to get version 1 of the feature ready for production. The tickets created for the issues mentioned here are https://phabricator.wikimedia.org/T194449 and https://phabricator.wikimedia.org/T194448.
Hi @קיפודנחש, thank you very much for the encouraging feedback :-) Regarding your suggestion: Generally, we wanted to provide an interface that is not overloaded to make it easy to use. Especially representing both + and - in a way that the interface stays clean is quite tricky, though it sounds like a use case people would often have. During the beta feature phase, we hope to learn more about the searches people are especially interested in, and we might look into supporting negatives later in the course of the beta phase. Nethertheless, you can still use all specific search options by typing them manually into the search field - I hope that helps for now! Best, Birgit
Not sure I follow this line of reasoning. The negation operators (-, ! and NOT) already exist and work with most search terms. Help:CirrusSearch says Truth-logic understands - or ! prefixed to a term to invert the usual meaning of the term from "match" to "exclude". and later A filter can have multiple instances, and negated instances. I would argue that making these operators work for deepcat and hastemplate makes search less complex, not more. There's less cognitive overhead in learning "I can put - before a search term to invert it" vs. "I can put - before most search terms to invert them, except for terms using the deepcat, and hastemplate filters".
thanks. i've been using the manual typing thing for a while now, and will probably continue to do so. one down-side of doing it manually is you miss cool new un-advertised features, like "deepcat", which i wanted for the longest time, and was not aware it was added (was it advertised somewhere, when i wasn't looking?)
peace.
Hi @קיפודנחש, you're right, that helps a lot to make the existing options more visible. Out of the same reason, we wanted to have info boxes next to the search fields, so that people can learn about the search syntax, and can find the Cirrus search help page for more information. About deepcategory: It is a fairly new keyword and still work in progress, this is why it only got announced via the discovery mailinglist. I think it would make sense to advertise it more broadly once more work is done, or announce it via the AdvancedSearch updates. About the work in progress: Not all category trees in the different wikis are indexed yet. That means that the "pages in these categories" field works for deep category searches on enWP, heWP, deWP ... among others, thereas it works like "incategory" on others. Category trees of more wikis are planned to be indexed soon, so that they can be searched within the given limits in more wikis (a hint about that will be added to the infobox for the keyword on AdvancedSearch with next week's software update). Regarding supporting negatives: I filed phabricator tickets so that we have it on our radar: T194449 and T194448. Thanks again for the suggestion :-)
one confusing term people often use is "category tree". the categorytree extension is nice and useful, but really, there is nothing in wiki categories that stipulates a tree structure. i'm not talking about the fact that people use the ward "tree" when they really mean DAG ("directional acyclic graph"), but about the fact that it's not even a DAG: there is no reason to assume the category graph is acyclic. directed cycles can, and sometimes do exist in the category directional graph.
peace.
Just to clarify, I want to point out that, contrary to the original post, negation does work for the incategory: filter. For example, on enWP incategory:"Musicals" -incategory:"Lists of musicals" correctly returns the 5 articles which are in category "Musicals" but not in category "Lists of musicals". Not sure if original poster was mistaken about this, or if it's been fixed in the 2 years since.
Thanks. to clarify: the request was to add the negation (which, as you noted, exists in the search logic itself) to the "advanced-search" feature. the advanced search is a gui, or "menu driven" tool that lets you use dialog boxes and menus to build the search pattern. this interface does not support the "not" modifier.
IOW, click on the magnifying glass icon, and then open the "advance-search" drop down. this request is to add support for "not" logic to this interface.
i think your statement "contrary to the original post", points to misunderstanding of the original post.
(btw: if you read the original request, you'll note that i asked, in addition to the "not" logic, to also add the "hastemplate" magic-search-phrase. note that this part of the request _was_ fulfilled, i believe long time ago).
peace.
Ah, okay, yeah, in that case I totally misunderstood this post. I encountered an issue where negating a deepcat query (directly via the query syntax) wasn't working, and came across this thread when searching phabricator for related issues. Though after some more experimentation, it seems to be less an issue of "negation isn't working with deepcat" and more "deepcat is generally broken". I filed task T238686 with some examples.