likewise all() just return all the docs if no callback specified
These features would make nstore easier to play w/ from the node-repl
It's a neat idea, but that means re-implementing all the logic twice, once using blocking I/O and once using non-blocking I/O. Not that I'm against the extra work, just not sure it's worth it. I want to keep nStore as simple as possible.
Also there is a strong feeling in node itself that we shouldn't encourage people to use blocking I/O. If I have an easy to use blocking API people won't realize the danger of it and just use it because it's easier. Then they will wonder why their server slows to a crawl when using nStore and blame me.
I'm not against the idea, just want to make sure you know the costs. Do you still think it's a good idea? Usability from the repl sure would be nice.
If it is that much work, I don't think it is so critical, esp. in light of other priorities
such as your need to focus on particular oss projects. I think Connect deserves the most attention ;)
If you feel comfortable hacking on nStore you can implement sync versions of stuff, just make sure to test it like crazy. data stores need to be super solid.