Michael Pollmeier

We just [retired](https://github.com/ShiftLeftSecurity/overflowdb/pull/288) the tinkerpop wrappers to free up the path for a new database backend implementation. If there's demand and/or other contributors/maintainers, we may bring it back later. Would...

Hi Jeremy, I'm glad to hear that you're interested in OverflowDB! As you probably noticed, the docs aren't up to scratch, and that's actually intentional - we don't consider it...

First of all: this is a really big improvement to the status quo, well done and thank you! I repeated your benchmarks by directly comparing only the graph operations, leaving...

For my understanding: we would still use the public API of our generated nodes/edges, but compared to of NodeRef/NodeDb they would not prevail in memory all the time (ignoring the...

👍🏻 that all sounds great! I envisaged that `NodeDb` will go away, and yes it makes sense to just make the generated nodes subclass `NodeHandle`, which is essentially a `NodeRef`...

I'm on board - let's move forward with this! @fabsx00 @ml86 any thoughts? @bbrehm what's the next steps? Do you want to continue this (e.g. add edge properties, property indices)...

> @mpollmeier What do you think about using a java/scala mix for odb-core? Yup I also it's time to bring scala into core as well - Denis and I were...

Yes, that's what I would expect. There's currently no measures of any kind in place to guarantee read/write consistency and atomicity. Almost everything is optimized for fast reads. We're working...

Do you really think that `String.intern()` is on the hot path for mesh? I very much doubt it... Apart from mesh being closed source, the entire stack is dockerised. So...

> i would also want to see some benchmarks before continuing with this. would you be happy with the one described in my comment above?