Things to consider before updgrading to feed-io 4.8

[alexdebril]

Hi,

I just wanted to inform you that the last feed-io version handles feed's "lastModified" date a bit differently than before.

Here's the thing: I've recently spotted some funky feeds which are exposing a "lastBuildDate" at the feed's level that is older than the newest item. As a consequence, feed-io returned items already consumed, leading to duplicate entries in database. This problem is solved thanks to a hack in 4.8, Here's the diff: alexdebril/feed-io@b2eb170 . As you can see, I don't trust the feed anymore and update its "lastModified" attribute if an item is fresher than the feed itself.

It works fine for https://api.feed-io.net/ and I think it should be also the case for nextcloud. If it doesn't, please rollback to 4.7 and ping me so I can look for a solution.

Also, I'm more and more convinced that internet has too many weird feeds for us to keep consuming them in a "clean" way (by "clean" I mean taking care of the HTTP headers, evict old items and so on). The dirty but safe way is to always download the entire feed and filter its content by querying the database in order to ignore duplicate items. It's gross and it wastes resources, but it makes users happy.

Maybe a trade-off for NextCloud would be to provide a "dirty-mode" that users could turn on for specific feeds. Default would be to read it "cleanly" (the actual logic) and for "bad" feeds (like Google's web.dev), the user switches the "dirty-mode" on to have it running.

[Grotax]

Thanks @alexdebril for the update I already spotted this change, I'm sure we will receive some feedback if feeds break.

Regarding a "dirty" mode I'm not sure but I will think about it.

[Grotax]

Oh and it's really cool to work together with you :)

[anoymouserver]

ping @alexdebril for #1109

[alexdebril]

Hi @anoymouserver , I've just had a look and answered. It looks like the user messed around with the files' privileges :)