search

snarfed / bridgy

📣 Connects your web site to social media. Likes, retweets, mentions, cross-posting, and more...
https://brid.gy
Creative Commons Zero v1.0 Universal
710 stars 52 forks source link

original post discovery: handle sites with high post velocity better #771

Open snarfed opened 6 years ago

snarfed commented 6 years ago

right now, we fetch a site's h-feed(s) and individual h-entries to find syndication links. we do this every 6h if we've seen a synd link recently, ie in the last 2w, otherwise we do it every 2d.

as a hardening measure, we stop after the first 10 entries per fetch. background in #624 and c9e30a2. this means we can miss posts on high velocity site's like @aaronpk's. we currently fetch and process / + /all + /syndicated on his site, which often totals 10-20 or more posts per day. anecdotal post counts per day recently: 13, 13, 25, 22, 25. IRC discussion.

one option is to bump up the 10 post per fetch cap, either globally or for known high velocity sites. another option is to let those sites tell bridgy which feed to use, e.g. just /syndicated, either via UI or rel value or other markup hint.

snarfed commented 6 years ago

bumped the per-crawl h-entry cap up from 10 to 50, just for beta users (2b47c069e6ffedb8251c0918d1b541f15288db65). also fixed a bug where manually triggered crawls like Crawl now and Resend weren't updating the last syndication url timestamp.