Skip to content
This repository has been archived by the owner on Mar 21, 2018. It is now read-only.

Need a rough frecency plan #1360

Open
jsantell opened this issue Oct 8, 2016 · 1 comment
Open

Need a rough frecency plan #1360

jsantell opened this issue Oct 8, 2016 · 1 comment

Comments

@jsantell
Copy link
Contributor

jsantell commented Oct 8, 2016

Overall, Tofino's usable for me for day to day browsing. One thing that currently is breaking my flow is the autocomplete frecency. On a fresh profile, I went to type in a subreddit, and typed it wrong:

reddit.com/r/somesubredditwrong

Then typed it again, correctly

reddit.com/r/somesubredditcorrect

Now when typing in this url, I get this ordering of results:

reddit.com/r/somesubredditwrong
reddit.com/r/somesubredditcorrect
reddit.com/r/somesubredditcorrect/blhablah
reddit.com/r/somesubredditcorrect/blhablah2

I'll always have that 'wrong' result up top, even though I only went there once, on accident. We should sort by visit, and maybe do some URL chunking to prioritize paths that appear more common in visits. Maybe this is what Firefox does, I'm unfamiliar with the internals of what goes on there, but surely we can add some rudimentary ordering

@rnewman
Copy link
Contributor

rnewman commented Oct 9, 2016

Desktop doesn't do much:

screenshot 2016-10-08 17 29 09

It sorts history and bookmarks by frecency for the list (which is why the wrong one is at the top for me — I just visited it). In the edit box itself it offers autocompletes for domain, then once you start typing paths, it offers shortest matching subpath for autocomplete:

r => reddit.com
reddit.com/r => reddit.com/r/
reddit.com/r/s => reddit.com/r/somesubreddit/

On mobile platforms we offer a sanitized top 100 domains list for autocomplete on an empty browser, so typing "fa" on a clean install offers you "facebook.com".

A prerequisite for this ticket is computing, maintaining, and persisting frecency in the UAS. This is a great opportunity to test our hypothesis about modeling these derived properties in a CQRS-ey fashion, using materialized storage and change observers. This depends on https://github.com/ncalexan/datomish/pull/14, which isn't far from landing; once @ncalexan gives me a status update on that, I'd be happy to take a look at this.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
Development

No branches or pull requests

4 participants