784653 - Don't offer reader mode on forum/comment-like pages (Bugzilla, Hacker News, etc)

Status: NEW

(Toolkit :: Reader Mode, defect, P4)

Description

[Lucas Rocha (:lucasr)]

We should only target article-like pages for the reader.

Comment 2

[Aaron Train [:aaronmt]]

Killing that P2 priority, probably doesn't stand.

Comment 3

[08xjcec48]

Add Github to the blacklist. As you can see on this page, only a single comment is displayed if you click on the Reader View button: https://github.com/gorhill/uBlock/issues/806

Comment 4

[Alfred Kayser]

Note, also the "Add to readinglist" icon should not be shown for these kind of "non-article" pages.

Comment 5

[:Margaret Leibovic]

(In reply to Alfred Kayser from comment #4)
> Note, also the "Add to readinglist" icon should not be shown for these kind
> of "non-article" pages.

The reading list and reader view are separate ideas - we support adding any kind of page to the reading list.

Comment 6

[Justin Dolske [:Dolske]]

Github is actually working reasonably well with recent fixes. Hacker News comments can trigger our simple heuristics (the icon shows up), but Readability barfs on the page.

I'd suggest we take this bug on one of two directions, or else it's impossible to ever close:

1) Split the bug into specific broken sites and deal with them individually. (Perhaps by heuristic improvements, or a blacklist to exempt them from reader view.)

2) Improve our simple heuristics to filter out more of these sites. The code landed in bug 1143844 is essentially looking for "5 paragraphs of sufficient length", but maybe it should (strawman) subtract a small penalty when it encounters a too-short paragraph. That would help with forum/comment threads where there are lots of short comments, and a few wordy ones.

I suppose there's also a (3)rd option, to make sure these pages are at least minimally usable in reader view. It's ultimately up to the user to use reader view or not, with lots of grey areas if reader view makes sense, but it should at least present a reasonable view of the page. EG, some forums are infamous for being filled with lots of junk (ads, GIFs, signatures, etc), and wanting a reader view is not entirely unreasonable. But this feels like a pretty low-value route to take.