Bug: Filtered decks are possibly not respecting easy days settings

Anki Help
1

I started using filtered decks for most of my reviews recently. Before that, I used to have very few reviews on Fridays (because I’ve changed it in Easy days settings to minimum). Now, the number of cards scheduled on Fridays according to statistics page is significantly higher. I thought that was because of Ankidroid, but they say that their code depends on Anki desktop for almost all backend operations, so I decided to report here.

Steps to reproduce the issue:

Requirement: you need FSRS parameters that schedule cards after first “Good” grade further than 2 or 3 days, to allow fuzz to work. Here is parameters from my test preset which reproduced the issue for me:

2.5778, 7.0101, 14.0491, 18.3035, 6.3987, 0.7538, 3.0185, 0.0701, 1.8277, 0.2010, 0.7579, 1.5691, 0.0443, 0.4324, 1.6635, 0.4273, 1.9097, 0.6590, 0.0424, 0.1298, 0.1420

Note: These are not my personal FSRS parameters, it’s just a test preset for test cards :sweat_smile:

  1. Create a test preset.
  2. Change learning steps to 1 or 0 steps to allow FSRS to schedule cards after first “Good” grade.
  3. Make sure there is no easy day.
  4. Make 10 test cards (name them from Test 1 to Test 10)
  5. Answer them all “Good”, but count how many of the 10 landed in each day. (i.e. how many landed after 15 days, how many landed after 14 days …etc)
  6. Undo all answers
  7. Pick the day where most cards landed, make it minimum.
  8. Answer them all “Good” again, you should observe that the number of cards that landed in the minimum day is decreased.
  9. Undo all answers again.
  10. Change Easy days again to be all normal.
  11. Now make a filtered deck with filter deck:"test". Make the order of cards by order added.
  12. Answer them all “Good” them again and count.
  13. Undo all answers again.
  14. Pick the day where most cards landed and make it minimum.
  15. Answer them all “Good” again. you should observe that the number of cards that landed in the minimum day is the same.

Expected behaviour
Filtered decks should follow all applicable preset settings (e.g. easy days, leech threshold (which I did not test)).

2

You say filtered decks aren’t respecting easy day settings. But why should they do that? Filtered decks should pull in all the cards that match the search criteria, which means that no limits other than those in the filtered deck creation dialog should be applied.

E.g. if friday is your easy day but you create a filtered deck on friday that pulls a lot of cards, then that’s expected behavior. There shouldn’t be any additional cap, just because of easy days.

Unless you mean that if cards are being reviewed with filtered decks and you grade the card, then the interval is falsly calculated without easy day settings applied.

3

Yes that is what I mean. It is a bug, which can be reproduced with the above steps.