Could the settings in the Stats window be sticky, especially the first row deck/versus collection - I end up flipping it every time I open up the stats. Might as well make all the other settings sticky in this view as well. Thx.
On large collections, it can take much longer to render the graphs when all time+whole collection is selected, so I’m a bit worried that persisting the selection will result in longer loading times. The current approach’s rationale is to show short term data quickly, and the user can then opt into the wider scope without it affecting initial load speed the next time they open the graphs.
So far the tests/usage here with my 2017 MacBookPro shows no difference really with the speed for showing different values (at least when switching). If this is an issue, maybe a hidden or visible preference then would help. I think the majority of at least Mac users by now have 2016+ models due to the depreciation of HW from Cupertino.
Or if nothing else, at least the current deck versus collection (maybe should be called all decks), I flip that over basically 99 of 100 times myself.
As Damien said, it depends on the collection. If you don’t have that many reviews, you would probably not notice a difference even with slow hardware. On my large test collection, it takes about four seconds to load the stats page and there’s nothing to look at in the meantime, whereas you already have some pretty graphs when you only switch to the slow settings later.
That being said, if the user chose the slow settings deliberately before, I think the delay would be within reason and he could be expected to make the connection and toggle the settings himself if the delay is unacceptable.
But the custom filter should even more be saved as it requires more work from the user to restore and shouldn’t be as slow as the whole collection option.
The term collection is used consistently in Anki and not always replacable with all decks.
Another option is to make the stats sections configurable, checkbox with no-need and activated will run and show stats. There are parts I never look at such as hourly breakdown and answer buttons. If that wil speed up the overall rendering…
Rumo, is the 4 seconds for the collection scope, or all history? I’m wondering if maybe we can compromise and have it remember deck/collection but always default to 1 year.
My sample collection is just 1 million cards without any reviews, so the period of time doesn’t make any difference. For my personal collection, the difference is just noticable.
Do you have anything better for benchmarking? If not, it’d probably be worthwhile to set up some tools to generate sample collections.
Anecdotally, collections above 200k cards are fairly rare, and heavy users tend to (but don’t always) have a higher review count than card count.
I guess we could have some other setting that controls whether these settings are persisted, though that’s a bit ugly. Or maybe we could just expose the defaults as a config var which people could use the debug console to adjust.