Executing “Ignore sentences” (that is, indicate that all sentences that have the same cloze word as the current sentence should be ignored) takes far too long (more than 60 seconds). I would like to do it much more often, but the long wait time discourages me. Would it be possible to do either or both of the following?
(1) Index the sentences by cloze word, then upload the index along with the sentences so that it will be downloaded with them. Then, when a request is made, it can use the index rather than sequentially searching through all the sentences in the database.
(2) Make the operation of marking the sentences that are to be ignored asynchronous rather than synchronous. In other words, allow the user to continue using Clozemaster while the operation is underway (though you may not want to let the user to exit from Clozemaster until the operation is finished). I suppose there’s the possibility that they may encounter one of the sentences to be ignored before the system has a chance to mark it ignored, but that seems unlikely, and there are probably ways to deal with it.
I said that it took more than 60 seconds. This time, I measured more exactly and found that it took 2 minutes (give or take a second). The cloze word for which I was trying to remove sentences was Мама (“Mama” in Russian). What’s worse, the time before this, when I was trying to remove sentences with the cloze word Тома (that is, “Tom” in the accusative or genitive case), the dialog seemed to hang for 5 minutes or so, but because I knew that this operation took a long time in general, I was reluctant to cancel it.
Could you please consider the ideas I suggested earlier?
You could try creating your own collection of sentences (‘My Cloze-Collections’) from the tatoeba corpus.
You can upload a file, among other options:
File should be a TSV (tab-separated values) file with 3 columns: sentence, translation, and pronunciation.
The sentence can indicate the missing word by wrapping it with {{ }}, for example This is the {{missing}} word. If no missing word is provided, one will be automatically selected.
Anki style This is the {{c1::missing}} word. and Readlang style This is the [[missing]] word. are also accepted.
Thanks for letting me know about “My Cloze-Collections”. However, it’s a completely different feature, with a very different workflow, and using it wouldn’t address the slowness of the “Ignore sentences” operation within the standard workflow (which I otherwise like). @mike, do you have any thoughts about the problem?
@alanf_us yep agreed that’s annoying / too slow. Will move it to async as a quick fix. Downside is that if you “ignore all” and then immediately search for the cloze you might see a few that don’t come up as ignored. Already working on a faster way to ignore all for the next iteration of Clozemaster. Thanks for the feedback!