search

mafintosh / hyperdb

Distributed scalable database
MIT License
752 stars 75 forks source link

Is hyperdb still intended to be integrated into hyperdrive? #169

Open e-e-e opened 4 years ago

e-e-e commented 4 years ago

:wave: I notice that the prerelease version of hyperdrive is using hypertrie rather than hyperdb, what does this mean for the future of hyperdb in the dat ecosystem? Is it still being developed or is multiwriter dats going to be solved in another way?

Apologies for my ignorance, I have not been following the dat project for several months and am just now trying to get back up to speed on everything.

mafintosh commented 4 years ago

Hi @e-e-e 👋

I spec'ed out the Hyperdrive road map in the decent DTN talk I did in Berlin, https://youtu.be/8fxGUiuMkTU where I talk a bit about the motivation behind this.

Hyperdrive is actually backed by a new module called mountable-hypertrie, https://github.com/andrewosh/mountable-hypertrie which allows you to compose a single kv store out of many different ones (ie you can mount them inside each-other).

This is super cool as it unlock a series of great collaboration patterns which very little technical complexity, preserving (and boosting) our random-access read speeds etc.

Now to get back to your main question, we have a plan to implement what we internally call "overlapping mounts" down the line. That means multiple mounts that can be mounted on the same mount point to create a hyperdb data structure. This data structure will, down the line, form the guts of hyperdb (it will solve some deep technical stuff in a much cleaner way), but it's pretty far down the roadmap given our current resource availibility.

The main trickyness with o/ is the getting the UX right and it ... just takes a lot of work :)

We have a bunch of really nice goodies coming out first though.

eklem commented 4 years ago

Cool! Where should I put some financial support?