MobileRead Forums - View Single Post

chaley · 03-19-2016, 03:42 PM

I wonder whether the 36,500 number is some kind of calculation error from the toolkit. It (the toolkit) is doing a bunch of stuff to come up with a curve then partitioning it by users. With a small number of users, I can imagine it getting it wrong.

I am changing the data gathering so that I can get both an average and the distribution of the size across a set of "buckets". That should eliminate the possibility of computation error and it will give me what I need to know. I don't care at all about exact numbers, but can use rough sizes for performance choices.

And I agree, getting numbers from the larger installed base will provide clarity. I will make the production release in a day or so, releasing it first to the beta group. A day after that I will start the staged release to the rest of the world.

03-19-2016, 03:42 PM	#7
chaley Grand Sorcerer Posts: 12,461 Karma: 8025600 Join Date: Jan 2010 Location: Notts, England Device: Kobo Libra 2	I wonder whether the 36,500 number is some kind of calculation error from the toolkit. It (the toolkit) is doing a bunch of stuff to come up with a curve then partitioning it by users. With a small number of users, I can imagine it getting it wrong. I am changing the data gathering so that I can get both an average and the distribution of the size across a set of "buckets". That should eliminate the possibility of computation error and it will give me what I need to know. I don't care at all about exact numbers, but can use rough sizes for performance choices. And I agree, getting numbers from the larger installed base will provide clarity. I will make the production release in a day or so, releasing it first to the beta group. A day after that I will start the staged release to the rest of the world.