Thanks for that report. I will check the code of sorting.
I find that your card is in a filtered deck. What if you move it to a normal deck?
Edit: What’s the retrievability of this card displayed in the retrievability column of browser?
I first experienced the behavior when studying a normal deck. E.g., see my first screenshot.
The second screenshot uses a filtered deck because I had already reviewed the card, and had to revert to a backup to check the memory state using the FSRS helper add-on. So I created a filtered deck to rapidly fetch the card.
By the way, I re-tested with a normal deck:
EDIT: The retrievability column of the Browser shows a Retrievability of 90%.
OK. Here are two problems:
- The Due is incorrect. 2023-08-30 (your real last review) + 47 days = 2023-10-16. So 2024-11-17 is incorrect. But I don’t know what causes it. It desires more analysis.
- The retrievability is inconsistent between normal deck and filtered deck. I have located the problematic code. It will be fixed soon.
Edit:
Just a guess. Possibly, “Reschedule cards on change” adds the interval to the current date instead of adding it to the last review date.
Edit:
It’s not possible that “Reschedule cards on change” would have this bug in every case. Otherwise, people won’t get any backlog after rescheduling.
But, maybe this happens for some specific cases.
An update on timing: it will unfortunately be about a week until I can build a new RC with the latest fixes.
Could you let me know the steps to reproduce with your AnkiWeb collection?
Here are exported decks with such an issue: link1 link2
I’m not sure what exact steps created this situation. I’ve got 2 candidates:
- After migrating to FSRS5 I’ve ran rescheduling for all cards using FSRS4Anki helper add-on.
- I am using filtered decks to review ahead parts of the material. Cards that are currently filtered tend to show due date way in the past in stats. However, the issue persists even if I empty the filtered deck.
Curiously, no backlog shows when those files are imported. Could it be one of their cards is in a filtered deck or is missing from the export?
Here is a link to a whole collection export, without media: collection-20241118183101.colpkg - Google Drive
Cards with this issue are in the “jezyki obce::arabski” and “jezyki obce::francuski” decks.
I submitted a PR to solve it: Fix/FSRS progress sometime shows 0 reviews by L-M-Sherlock · Pull Request #3591 · ankitects/anki
I didn’t realize this deck had old review logs before I started it, but even then this cant be right can it? 4.28 years?
Apologies if this isn’t the right place for it:
On 24.11rc1 (MacOS 15.0.01) in attempting to conduct a one-way force sync:
I get a message saying that the sync is complete:
However, the ‘Sync’ button at the top is perma-red and does not appear to be doing much, despited repeatedly attempting to sync. Have disabled all add-ons and restarted Anki, to no avail.
Notably: the prompt to download or upload to Ankiweb never appeared, and syncing from a second device it does not appear that the one-way sync took place or was pushed.
Closing 24.11rc1 and opening 24.6.2 I was immediately greeted with the appropriate Ankiweb prompt and, after doing all the normal stuff, all appears to now be well. This did the trick, the one-way sync was pushed (and received by other devices), and the red sync button is now gone.
After doing that, reopening 24.11rc1 everything appears to be hunky-dory. No error messages, nothing. Do with this, ye Gods, what ye will
¯_(ツ)_/¯
In the view of FSRS, you even recalled the card after 1.5 years. So the interval is expected.
When I look at those two decks, they only have a backlog of a few days. I tested with both the current development code, and 24.11rc1. Are you on the latest build? Do you see it when importing the colpkg into a new profile?
I believe it’s a corner-case where a full-sync has been flagged, but the collection has otherwise been unmodified. Any trivial change to the collection should resolve it. The act of setting that flag should also set the collection as modified, and I can’t reproduce the issue here. Maybe your system clock went back in time.
I can report, happily, that on a lark I tried the same thing a second time and it went off without a hitch.
So potentially just some random fluke on my end, or with my machine, and maybe not a genuine bug of any kind.
I don’t see this issue in the latest build. Seems to be fixed somehow.
EDIT:
Definitely fixed. What I checked on my PC right now:
- Version 24.10rc1 - issue visible in stats.
- Update to 24.11rc1.
- Issue no longer exists.