On 10/25/2016 09:58 AM, Paul A. Rubin wrote: > On 10/25/2016 12:14 AM, Jürgen Spitzmüller wrote: >> I think the right approach to fix this is to use categories, like we >> have in layouts (the interface is already implemented also for >> modules). This would make it easier to find the module you are looking >> for (without knowing the exact name, which is sometimes quite >> arbitrary: did you know that there is a further theorem module sorted >> under "N"?). This would probably also save horizontal space. >> >> So instead of: >> >> ... >> Named Theorems >> ... >> Theorems >> Theorems (AMS) >> Theorems (AMS, Numbered by Type) >> Theorems (AMS-Extended) >> Theorems (AMS-Extended, Numbered by Type) >> Theorems (Numbered by Chapter) >> Theorems (Numbered by Type) >> Theorems (Numbered by Type within Chapters) >> Theorems (Numbered by Section) >> Theorems (Unnumbered) >> ... >> >> We would display: >> >> ... >> Theorems >> |- AMS >> |- AMS, num. by Type >> |- AMS, extended >> |- AMS, extended, num. by Type >> |- Standard >> |- Named >> |- Num. by Chapter >> |- Num. by Type >> |- Num. by Type within Chapters >> |- Num. by Section >> |- Unnumbered >> ... >> >> A filter bar to filter for categories and names (plus probably >> descriptions and/or keywords) would further increase usability. >> >> Jürgen >> > I have no objection to categories, but I suspect we will end up either > with a large number of categories containing one or two modules or a > single large "miscellaneous" category. > > With or without categories, I would definitely vote for the filter > bar, and I think it should include descriptions. With the filter bar, > Andrew could type in "Assumption" and find all variations of the > theorems module containing assumptions.
This assumes (!) that the Assumption style is mentioned in the description, which it might be. It would be possible, but a whole lot more complicated, to filter on the styles, etc, that are declared in the module (if any). Richard
