Unread status of archived articles when adding a new feed

What is the expected behaviour when adding a feed while subscribed to the archive tier with the “Mark read” set to 365 days?

This feed obviously has an archive, but I only got 35 articles as unread, the oldest is a month old. Perhaps this should honour the mark read (similar as to what seems to be happening to other feeds where older articles are returning)?

I’m not sure if there is any harm (resource consumption on NewsBlur’s side) to bringing old articles as unread, I expect in many/most cases I would end up just marking everything older as read anyway, but there are cases (this being one) where I wouldn’t mind having a longer history marked as unread as I am going through a bunch of older articles.

image

This is a good question. Unfortunately this answer won’t be that satisfying but it isn’t 100% predictable. NewsBlur defaults to using your mark as read date, which is either when you last had your unread count at 0 or literally hit mark as read. But hitting 0 doesn’t always change the mark_as_read date because there’s certain conditions where the count is stale so moving the mark as read date forward would cause you to lose newer unread stories that just havent come in yet.

I could fix that by having a check for the latest story date every time you hit 0 unreads, but then there’s hidden stories which stay unread but shouldn’t be considered in the movement of that mark read date. In fact I believe it’s already doing some of this, but looking through the code it’s clear that the mark as read date may move to the unread cutoff (30 days or 365 days or whatever) if it’s older than that.

All that is to say that it’s a moving target and getting that count right retroactively isn’t trivial.

But the good news is that it basically works correctly on new stories going forward.

Interesting, thanks. I really appreciate the insight into what it is doing internally (and why).

This also highlights the difference between some third party clients (Reeder) and your own, in that older Read articles never get synchronized at all.
Usually this is probably unnoticed, but I was hoping to (manually) “transfer” some items from another queue and Reeder is quicker.

No big deal, I can still do what I want. I’m a bigger fan of the archive tier than I had expected. Thanks!