Re: org-drill extremely slow with Org 9.2.5

[Milan Zamazal <pdm@zamazal.org>]

>>>>> "OK" == Oleh Krehel <ohwoeowho@gmail.com> writes:

    OK> I noticed org-drill being slow three years ago when I tried to
    OK> learn it.  So I wrote my own package:
    OK> https://github.com/abo-abo/pamparam/.  It's quite fast: it takes
    OK> 0.6s to sync my 3300 cards from the master Org file.  And
    OK> day-to-day learning operations like building a schedule or
    OK> fetching a card are instantaneous.  The master file is
    OK> relatively small, since it stores no metadata: less than 10000
    OK> lines.  The metadata is stored per-card, each card is in its own
    OK> file. The whole thing is backed by Git.  All your learning
    OK> sessions are stored in commits as well.

    OK> Check it out. It might have less features, but it's really fast
    OK> and has served me well.

Hi Oleh, pamparam looks interesting, however the very first blocker of
adoption is that having to type the answers is slower than anything
else...  And not always feasible.

Nevertheless the concepts behind pamparam are interesting.  When I grep
all the :PROPERTIES: out of my org-drill file, Org can visit and
navigate the resulting file completely fine.  So the problem is how Org
deals with its properties.

I can see two options:

a) to fix Org property processing
b) to change org-drill to off-load metadata to a separate file

I'm afraid b) alone wouldn't help, since when I retain just :ID:
properties, then the file is unusably slow in Org.