Hi,
I try to figure out what is the relationship between FSRS settings and “New cards/day”. Is there any? Are they independent? May it is under development? I ask to better understand how to manage my workload.
“Compute optimal retention” suggest that there is some relationship due to “Minutes study/day” but when I change “New cards/day” then it compute the same values.
Thanks
FSRS retention calculation doesn’t take other deck settings into consideration, only the values you provide in this section.
“New cards/day” works exactly the same way with and without FSRS, and it doesn’t affect the output of “Compute optimal retention”.
I know I said this before, but an option to reoptimize parameters after a certain number of days (or reviews) would be amazing. Right now, users have to remember to re-optimize parameters periodically, which is much less convenient than leaving it to Anki.
I agree. And could be based on the number of reviews someone does, not necessarily to a time period
I get this every time I try to activate FSRS:
DbError { info: “SqliteFailure(Error { code: Unknown, extended_code: 1 }, Some(\“Expression tree is too large (maximum depth 1000)\”))”, kind: Other }
anyone got any idea?
Anki doesn’t remember the preview window size, it did remember it in 2.1.65 but stopped in .66 and still isn’t fixed.
This is a big update, and it’s going to break some stuff. Addons that were using pointVersion() instead of doing their own version parsing will be unaffected.
This is not an important question, but why not “Anki 2.2.10”, “Anki 2.3.10” or “Anki 3.0”?
The logic is year.month (and eventually year.month.patch)
So 23.10 stands for october 2023.
This way, it becomes easier to identify when a version was published, and to distinguish at a glance between minor and major releases.
Thanks, that’s efficient.
After disabling FSRS, when I click on Save I get this error message and Anki crashes.
No add-ons active.
A fatal error occurred, and Anki must close. Please report this message on the forums.
Anki 23.10 (02ade8e7) Python 3.9.15 Qt 6.5.2 PyQt 6.5.2
Platform: Windows-10-10.0.19045
Flags: frz=True ao=False sv=?
Add-ons, last update check: 2023-10-09 07:51:30
Caught exception:
Traceback (most recent call last):
File "aqt.progress", line 118, in handler
File "aqt.taskman", line 126, in <lambda>
File "aqt.operations", line 122, in wrapped_done
File "concurrent.futures._base", line 439, in result
File "concurrent.futures._base", line 391, in __get_result
File "concurrent.futures.thread", line 58, in run
File "aqt.operations", line 107, in wrapped_op
File "aqt.operations.deck", line 88, in <lambda>
File "anki.decks", line 297, in update_deck_configs
File "anki._backend_generated", line 1557, in update_deck_configs_raw
File "anki._backend", line 151, in _run_command
pyo3_runtime.PanicException: command requires weights to be set on creation
Since there was no other reply, I think this bug seems to have been fixed in this commit: Fix 'Expression tree is too large' when user has many decks · ankitects/anki@0e6104a · GitHub. So this will be fixed in the next beta release if you don’t want to compile the source. I didn’t try yet to read the code and understand the nature of the bug, but judging from the commit title it has to do with the number of decks. So if you can’t wait for the next beta and don’t want to compile from source you might try a workaround of using fewer decks per preset (e.g. assigning preset copies to decks). Though I didn’t test whether that would work in practice.
Integrated FSRS not apply parameters “per preset” ??? 
I think this is a big problem for my type of decks. 
For me it was too. But I adapted. See if your situation is for the same reason as mine.
Here is the proposed solution.
If you switch the integrated FSRS on, it is activated for all presets, but parameters are set for each preset independently. First, you cannot use V3 for one deck and FSRS for another deck. You must use either V3 or FSRS. It is actually the same logic as there has been V2 and V3 scheduler. Second, If you switch FSRS on, you need to calculate weights for every preset.
I have tested FSRS a bit. I changed presets, set parameters, answered few cards and finally I wanted to switch FSRS off. I have got the message as below after switching off the FSRS toggle.
A fatal error occurred, and Anki must close. Please report this message on the forums.
Anki 23.10 (02ade8e7) Python 3.9.15 Qt 6.5.2 PyQt 6.5.2
Platform: Windows-10-10.0.22621
Flags: frz=True ao=True sv=?
Add-ons, last update check: 2023-10-09 21:04:09Caught exception:
Traceback (most recent call last):
File “aqt.progress”, line 118, in handler
File “aqt.taskman”, line 126, in
File “aqt.operations”, line 122, in wrapped_done
File “concurrent.futures._base”, line 439, in result
File “concurrent.futures._base”, line 391, in __get_result
File “concurrent.futures.thread”, line 58, in run
File “aqt.operations”, line 107, in wrapped_op
File “aqt.operations.deck”, line 88, in
File “anki.decks”, line 297, in update_deck_configs
File “anki._backend_generated”, line 1557, in update_deck_configs_raw
File “anki._backend”, line 151, in _run_command
pyo3_runtime.PanicException: command requires weights to be set on creation
Regarding the re-optimizing discussion, the main reason why I find it annoying is that I have to go through each preset and optimize it individually. Therefore I would also like to have either an automatic periodic optimization or a “Optimize all presets” option. With the latter I would only need one click, so it would be less annoying to do that periodically.
As optimization takes a while there should be considerations whether that can run in the background. And some thought should be put into how to treat errors during automatic optimization. So I think this is a good feature for future releases, but I have doubts that we’ll get that for 23.10.
Ah, ok.
Using weights per preset is enough for me.
About “Loss aversion”. Right now, I don’t see any way to understand what it does without:
a) Asking the devs or LMSherlock
b) Looking into the source code
It’s safe to assume that the average person will have no idea what it does. So either it needs to be removed or there should be a tip that appears when you click on it. Of course, that would be helpful for all FSRS-related settings, it’s just that this is the least intuitive one, it would benefit from having a tip the most. But ideally, it should be hidden. FSRS already has problems with being intuitive and easy to understand, “Loss aversion” is straddling the line between “hard to understand” and “you literally need to be a developer to understand”.
