The ‘id=answer’ part tells Anki where the divider is between the question and the answer. This allows Anki to automatically scroll to the spot where the answer starts when you press ‘show answer’ on a long card (especially useful on mobile devices with small screens).
It doesn’t mention any side effects of rearranging field positions.
In any case, I didn’t use {{FrontSide}}, so I’m a bit confused.
The Flip button in any case should be able to work with or without the quotation marks. Should be a simple fix, no? The version with quotation marks is standard HTML syntax.
Thanks for the tip about Find and Replace in Notetypes. It needed some simple modifications to remove Qt5 dependencies, so I sent in a PR to the author’s Github.
Because the simulation is inherently random, the output of the simulator will be noisy. To alleviate that, in the current implementation, the simulator runs multiple times during every iteration, and then the results are averaged. The minimization function (Brent or the old one, it doesn’t matter) works with the averaged output.
In the current implementation, it’s an average of 5 runs, aka sample size is 5. In other words, during each iteration, the simulator simulates the review history five times. Increasing this number will make the output less noisy, but it will also make minimization slower. Unfortunately, we can’t have both high speed and no noise.
You can read this code if you want to see the details.
Sorry to monopolize the conversation, but there’s some additional info to report.
When I actually tried to use 23.12beta1 to review one of my real decks (not the 1-card deck I made for the above tests), it gave the same kind of weird database error that some others have been reporting:
The above happened when clicking the Rebuild button on a filtered deck.
Check Database couldn’t fix it, and forcing a one-way sync to download a good version of the collection also didn’t fix it, because the same error kept happening when clicking Rebuild. Reinstalling 23.10.1 did work.
The bug I reported earlier in this thread was happening on 23.10.1, so I downloaded 23.12.beta1 to see if it still occurred. To create the simple two-field and one-card example, I cloned one of my real notetypes and then systematically deleted most of the fields and cards to simplify as much as possible. In the course of that process, I saw the bug that some others reported, namely the {{First Field}} magically getting appended after the end of an “if” block that encompassed the entire front template, as in:
{{#Meaning}}
...
{{/Meaning}}
{{First Field}}
I don’t know if all that deleting screwed up the database somehow.
But in the end, I only used 23.12.beta1 for a very brief time, creating and testing the test case card for the bug I reported earlier, and when I then tried to use it for real usage, it gave the SqlInputError bug above.
I just tested a bit more, and it seems that if there is a type field above <hr id=answer> in the template, the type field and everything between it and the hr will be moved below <hr id=answer>.
E.g., using an inspector, we can see that both {{type:Back}} and <p>example 2</p> are automatically ‘moved’ below <hr id=answer>
By the way, {{FrontSide}} is not necessary to trigger this. Simply, it is useful to have this kind of behavior when using {{FrontSide}}, because otherwise the answer (or at least part of it) would be displayed above the hr.
Conversely, if you are not using {{FrontSide}} in your template, this becomes less useful and may even have unintended side effects.
Talking about side effects, it seems that if the type field is placed below <hr id=answer> everything between the two will be ‘moved’ above the hr, e.g.
→ More probably and more simply, it seems that <hr id=answer> will always try to place itself right before <code id="typeans">
Yeah, at least in theory it sounds like it! I was mentioning the Flip button mostly to inform you about the differences in the way Anki currently handles <hr id=answer> vs <hr id="answer">.
Also, in my opinion, the “Optimize All Presets” feature is very inconspicuous. I would’ve never discovered it if I hadn’t read the changelog. I suggest moving it to the “Advanced” section, under “Optimize FSRS parameters”.
I also find the wording confusing, your suggestion, IMO is much clearer (also, in Ankidroid, the feature is under general preferences > review, which I much prefer). However, my main problem is that I can’t get Anki to automatically press “Again” or (bury card) and show the next card at all. It just keeps waiting manual input from the user, and afterwards, the “More > Autoadvance” option gets deactivated. Tried on my Windows and Linux machines, same behaviour.
Just after the auto advance is activated/deactivated successfully by the shortcut (Shift+A) or selecting the item in the More menu, it would be desirable to show its visual feedback (a popup), like Auto advance activated / Auto advance deactivated, I think.
Currently, users may feel somewhat uncertain and uneasy about whether the switching of the auto advance has been done successfully.
It says the bug with sorting by Retrievability (Relative Overdueness) is fixed. Yet I tried this, and normal decks are showing one result by this sorting on one hand, on the other hand, using the same cards in a Filtered Deck shows a completely different sorting.
Is this bug still not fixed because my other Anki client, Ankidroid, still did not fix this bug?
Yes, evertyhing working OK when using the shortcut, thanks!
Since you can disable this feature directly from the deck options (by setting “Seconds to show question” and “Second to show answer” to 0), isn’t it a bit redundant the extra command added to the “More” button?
@dae my suggestions ended up being scattered across this thread, so I’ll put them all into a single comment here:
The “Optimize All Presets” feature is very easy to miss. I would’ve never discovered it if I hadn’t read the changelog. I suggest moving it to the “Advanced” section, under “Optimize FSRS parameters”.
Sherlock finished benchmarking FSRS v4 on new data, ask him to share the new default parameters.
I agree with other users that “Seconds to show question” and “Seconds to show answer” are a little confusing. “Seconds until answer is shown” and “Seconds until next question is shown” are better.
Since the optimal retention estimator is now ~2 times faster, we could sacrifice some of that speed in favor of accuracy by increasing n in sample(config, weights, w, 5) to 7-10.
After further testing, I’m getting some unexpected results when using subdecks with mixed autoavance options, at least in some cases (Anki is automatically buring review cards belonging to subdecks where Seconds to show question” and “Second to show answer” are set to 0)
B. Many times the MPV process remains even after closing anki (especially if I close through the console X) this was also in previous versions and not necessarily in the new one.
It says the bug with sorting by Retrievability (Relative Overdueness) is fixed. Yet I tried this, and normal decks are showing one result by this sorting on one hand, on the other hand, using the same cards in a Filtered Deck shows a completely different sorting.
Is this bug still not fixed because my other Anki client, Ankidroid, still did not fix this bug? @L.M.Sherlock@dae
Ok it seems there’s some kind of incompatibility between the auto advance feature and the built in TTS.
This is what happens when using auto advance with a cards using TTS fields:
To replicate:
Create a test deck with some cards using Basic notetype
Add {{tts xx_XX:Front}} to the front or the back of the card
Deck options > Timer
Seconds to show question: 5
Seconds to show answer: 5
Answer action: Answer Again
Study the deck. Activate Autoadvance (Shift + A). Press Space bar. Anki will automatically press “Again” on all cards inmediately, as shown in the gif.
Now edit notetype, delete the TTS field > Everything works as expected.
