Call me old fashioned, but I firmly believe that two searches of the same website for:
* the exact same search terms
* in the exact same order and
* with the exact same formatting
in the same session of the same browser tab, separated only by a few seconds of time, should not produce:
* different sets of results or
* sets of results sorted in different, apparently random orders.
Like, how do you even manage to do that! The search and sorting algorithms you'd have to write are either the polar opposite of simple and speedy, incorporate a RNG, or both, all three of which would earn an F in any CompSci class.
@arielmt (contextualizing, of course, that the iconolog's body of work is not at all suitable for text searching and this was a band-aid to make it work a little better with it)
@mavica_again That's a very good and appropriately exceptional use case.
@arielmt could be streaming from a dataset that is constantly in flux, which considering regular spiders and algorithmic tuning doesn't seem too unbelievable
@arielmt RNG is precisely how i do it with @\iconolog in an attempt that you don't get the same icon twice for the same search: i do a fuzzy match for filenames and categories from your query, then shuffle the top matches above a certain threshold