I'm afraid that for me as a non-native English speaker (to say the least) it would be a little suboptimal to write the perfect documentation, unfortunately.
And I agree with you in that all the relevant pieces of information are there, and for me personally, as well as for saurik, source code is more than enough actually, especially considering they are really clear and compact.
I just wish LMDB were more popular and widely adopted because it's really brilliant piece of software. And I think documentation is a one of main factors in that, that's it.
Anyway, thanks for pointing, I'm thinking about a potential complexity of adding to LMDB the ability to use secondary indexes (separetely stored, due the carefully done tuning of main storage scheme) so I'm sure it'll come to discuss with you on IRC and mailing list.
Yeah, I've focused my attention more on the code than the docs, which is pretty typical for me. But frankly I think the doxygen stuff is already very illuminating.
For secondary indexing, you should take a look at what slapd back-mdb does. I know that BDB supported 2ndary as a built in feature but we never used it because it was too slow. What we're doing now is already quite fast.
And I agree with you in that all the relevant pieces of information are there, and for me personally, as well as for saurik, source code is more than enough actually, especially considering they are really clear and compact.
I just wish LMDB were more popular and widely adopted because it's really brilliant piece of software. And I think documentation is a one of main factors in that, that's it.
Anyway, thanks for pointing, I'm thinking about a potential complexity of adding to LMDB the ability to use secondary indexes (separetely stored, due the carefully done tuning of main storage scheme) so I'm sure it'll come to discuss with you on IRC and mailing list.