Non-default learning always falls back to the default order: Difference between revisions

From SuperMemopedia
Jump to navigation Jump to search
(Status update for new version)
Line 31: Line 31:
== Version update ==
== Version update ==
This bug is still present in 17.21.
This bug is still present in 17.21.
=== Tasklist priority ===
This job is still on SuperMemo 17 tasklist, however, the present update had to be rushed for release due to some bugs in Algorithm SM-17 (e.g. anti-cramming notes). Handling learning mode is surprisingly complex. The user only needs to know '''Learn''', but in the background, SuperMemo needs to consider dozens of scenarios in different combinations. This mode has been reworked in SuperMemo 17 for the sake of ''neural learning'', and caused a lot of debugging issues in beta-testing. In other words, this is not a task that can be handled in a hurry. There will be more testing needed by users using different strategies (other than just plain learning, or subset review).

Revision as of 10:59, 20 June 2017

problem

Gutmanas wrote:

I have a bunch of non-memorized elements in my collection.

When I right-click a folder in the contents window and select learn, I am shown the first element from that folder, as expected.

But, once I answer it, the next element shown is the next element in the pending queue - which isn't even from that folder!

I would have expected to see the next element in the folder instead.

But, if I press cancel and then learn again, I get a dialog asking whether to resume subset learning - and if I accept, I am shown the next element in the folder. But, the one after that is going to be the first in the pending queue again, requiring me to press cancel and learn again...

Note that I get the same behavior if I pick process branch > learning > learn on the folder rather than just learn.

A related bug is that if I open the pending queue, sort it by the order of the knowledge tree and select learn the displayed elements in the displayed sequence, my order gets completely ignored and learning proceeds according to the default order of the pending queue.

How can I workaround these bugs in the meanwhile?

It seems like subset learning works only for already memorized items, rather than pending items.

Answer

If you have a branch that you want to memorize in a specific order, the best approach is to Spread the elements over a period of time while retaining the repetition sequence. For example, you can spread the branch at 3 elements per day over a period. If you happen to have time for more than 3 elements on a given day, you can use Add to outstanding to add more elements of that branch to the outstanding queue.

pending queue problem

As for subset review and pending items, this would need to be investigated. Bugs are possible due to the fact that the pending queue is only used with ready-made collections. In newer SuperMemos, all new elements are automatically added to the learning process.

Version update

This bug is still present in 17.21.

Tasklist priority

This job is still on SuperMemo 17 tasklist, however, the present update had to be rushed for release due to some bugs in Algorithm SM-17 (e.g. anti-cramming notes). Handling learning mode is surprisingly complex. The user only needs to know Learn, but in the background, SuperMemo needs to consider dozens of scenarios in different combinations. This mode has been reworked in SuperMemo 17 for the sake of neural learning, and caused a lot of debugging issues in beta-testing. In other words, this is not a task that can be handled in a hurry. There will be more testing needed by users using different strategies (other than just plain learning, or subset review).