Anki 23.10 Beta 1-4

[pavreh]

@dae

Qt 6.5.1 changed the default driver to Direct3D. Presumably this works better for some people, but if it causes regressions for you, there’s now an option in the preferences screen (as you’ve discovered) to switch back to the older OpenGL option.

I experienced some glitches with both, Direct3D and OpenGL. Vulkan is a video driver which works for me.

[brubsby]

It’s currently unclear what cards the “Compute optimal retention” section is using for timing info. For instance, if I have “deck:All” in “Optimize FSRS weights”, but the preset I’m changing is a “parent deck” type preset that isn’t on any actual decks with cards, is it using the average completion time of all of the subdecks of “deck:all”? all the cards in subdecks of decks with the preset? Surely not. All the cards that are just in the exact decks with that preset? (but not their subdecks?)

I ask because there are situations where a Loss Aversion of 1.00 will get me a desired retention of 0.94, but a Loss Aversion of 2.50 will get me a desired retention of 0.92. Which only makes sense if: 1. My understanding of Loss Aversion is backward. 2. There is a bug in Loss Aversion. 3. it’s using the timings from the preset (or current top level deck only) and not the subdecks, where I have maybe 1 or 2 accidental cards that have noisy timing data.

[pavreh]

If I do other things on my computer while Anki calculates optimal weights, sometimes the GUI stops responding. It looks like frozen but after some time, when the calculation is finished, the GUI starts to respond again. It is a combination of a large collection (234,000 reviews) and an old computer.

[pavreh]

As I understand FSRS, the weights calculation is an important part of it. Then, “calculate” button shouldn’t be hidden in the settings. It should be visible right below the FSRS toggle.

[tatsumoto]

Anki 23.10 Beta

While it’s not too late, it’s probably better to call the next version 2.1.67 because the last version was 2.1.66. I bet there are addons that rely on aqt.appVersion and some of them might expect the first two numbers to be 2 and 1.

[Expertium]

Speaking of which, I wonder why the version naming convention has been suddenly changed.

[Aleksej]

How good is the optimization in Anki, which takes little on a CPU, compared to the optimization on Colab, which takes longer on a GPU?

[Spartandrew18]

Because I believe it’s more conventional and a lot of projects move to this type of naming scheme. Just my assumption.

[pavreh]

There is some info about versioning:

New toolkit and packaging test (Windows)

Anki has a long history of not-so-great version numbers. Semver is a bit of a tricky subject. From the perspective of add-ons, it’s a breaking API change, and you could argue that should make it 3.0. But that’s not really practical - because add-ons are so tightly coupled, we’d end up bumping the major version every few releases that way. So I mainly focus on what’s exposed to end users outside of add-ons. This update doesn’t change how cards are stored on disk, or require a down…

[dae]

@Expertium

I apologize if I’m being too annoying, but both in this thread and on Reddit people agree that FSRS doesn’t have to be complicated to us

This stuff is very new and still in beta. Even once a stable release is made, I suspect it will be some months before all the various kinks have been worked out. After that, we can potentially look at moving some of the FSRS options out of the advanced section, but I do not think the time to do so is now.

@brubsby

It’s currently unclear what cards the “Compute optimal retention” section is using for timing info.

It uses the same search as for computing weights.

@tatsumoto

While it’s not too late, it’s probably better to call the next version 2.1.67

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.

@Expertium

Speaking of which, I wonder why the version naming convention has been suddenly changed.

Use a different versioning system? SemVer, perhaps?

I see. Would love to see the change happen! Oh and by the way, I personally think year.month.patch would be a rather suitable versioning scheme for Anki as well.

@Aleksej

How good is the optimization in Anki, which takes little on a CPU, compared to the optimization on Colab, which takes longer on a GPU?

Achieve parity with the Python optimizer

[szalejot]

@Expertium

I apologize if I’m being too annoying, but both in this thread and on Reddit people agree that FSRS doesn’t have to be complicated to use, and that it’s entirely possible to make it accessible to new users if they only need to adjust desired retention. I believe that simplicity for new users should be top priority.

I second to this. I have a friends that were interested in app for memory learning, but didn’t want to use Anki, because it was too complicated at the start.
Also, the plain grey UI is not appealing to the users, that are used to modern nice UIs. But that’s a topic for a different discussion.

[Expertium]

Is there a reason to keep the “Custom scheduling” field when FSRS is enabled?

I think that when built-in FSRS is enabled, this field should be hidden.

[szalejot]

@Expertium

Is there a reason to keep the “Custom scheduling” field when FSRS is enabled?

You can still add some custom scheduling rules, on top of FSRS. There are separate functionalities so it’s ok to keep them both.

[Expertium]

@szalejot

You can still add some custom scheduling rules, on top of FSRS.

Yes, but realistically speaking, how many people will do that?

[Sqrt-1]

The option to do so should be there; what if someone needs to test out some extension to FSRS which is too unstable to be deployed directly into Anki?

[Expertium]

This is why I think that having two layouts/levels would be better, so that advanced settings and features would be available to those who need them, but not by default.

[Sqrt-1]

If I understand correctly, the current rationale for not doing that is that FSRS is very new and there are many kinks that need to be worked out. FSRS is intentionally placed in the advanced section so only advanced Anki users use it.

[brubsby]

@Expertium

Yes, but realistically speaking, how many people will do that?

In this very thread there has been examples of people adding custom rules on top of FSRS already for their own ends, I’m not sure why you feel such desire to make FSRS less flexible.

[Expertium]

@brubsby

In this very thread there has been examples of people adding custom rules on top of FSRS already for their own ends

Huh, I guess I just didn’t notice then. My bad.

@brubsby

I’m not sure why you feel such desire to make FSRS less flexible

It’s not that I’m against having certain features, it’s just that I think it’s better to prioritize simplicity. Just to clarify, I mean simplicity of usage, not simplicity of the algorithm.

[Shigeyuki]

I think under " Advanced" we need “Super Advanced” or " Extreme Advanced".