Peter Johnson

rebased `master`

linking https://github.com/pelias/pelias/issues/894 as it may be covered by this work

Thanks for the report, after some investigation we found that there is a conflict with the functionality which flattens English diacriticals such as 1st, 2nd, 3rd, 4th but we're still...

> Common German abbreviation are also not handled well 🟢 Heisenbergstraße 2, Münster 🟢 Heisenberg-straße 2, Münster 🟢 Heisenberg straße 2, Münster 🟢 Heisenbergstrasse 2, Münster 🟢 Heisenberg-strasse 2, Münster...

I also looked in to getting numerically ascending scoring in this case, such as: ``` ?text=stop 2 Stop 2 Stop 20 Stop 21 ... ``` ... however, this is not...

cc/ @fpurcell

I'm still interested in merging this, the 'Failures' represent a change, but I'm not convinced it's negative. With our current implementation, it's impossible to find some of those records, for...

I wrote up a wiki article explaining the pros/cons of different autocomplete query styles https://github.com/pelias/api/wiki/Autocomplete-search-logic

@orangejulius I had a look at the autocomplete queries today following on from writing the wiki article linked above ^ I think that it's appropriate to generate 4 different queries...

Also worth noting that our `min_ngram` size is currently set at `2`, so if a clause was to be generated against the `name.default` index which tested a single-character token, then...