Re: persistent data feature

[Matt Armstrong]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> But the current large data structures are in things like the Gnus
>> registry, which can grow arbitrarily large (and slow, unless you trim
>> it).  And my .ecompleterc is 4MB.
>
> 4MB doesn't sound like it should be a problem, and I don't see much
> benefit in having ecompleterc grow much further (by its nature it
> contains a lot of "irrelevant" data and it's good to trim it every once
> in a while for that reason).  Also having it too large would slow down
> completion itself, so I suspect that storing it in a sql database would
> not help.
>
> The gnus-registry is a much more compelling example (especially since
> there's no reason we should regularly scan it from start to end, we
> only really ever need to look up specific entries, so the indexing of
> the database should allow answering those queries efficiently without
> having to read the whole DB from the disk (tho nowadays reading 100MB
> from the disk is basically instantaneous anyway).
>
> So, IIRC the only current example is gnus-registry.  That sounds good.

Another example is Org Roam (https://www.orgroam.com/), which uses
sqlite to manage its index of Org metadata.