Closed kordano closed 3 years ago
A version of this by @alekcz is available here: https://github.com/alekcz/konserve-jdbc
should we close this issue? It might take some time until datahike-jdbc is ready to replace datahike-postgres
@TimoKramer I can get it there over the weekend. Do you have an idea of what's still missing for it to replace datahike-postgres?
Hi @alekcz and thanks! I think the main problem is that datahike-jdbc brings the konserve 0.6.0 with it and when you are using datahike it loads konserve 0.5.1 and then datahike-jdbc does not work. so this is basically a question of when we can use datahike with konserve 0.6.0 I guess. Or making konserve-jdbc work with 0.5.1?!
We need to make datahike work with konserve 0.6.0. We should probably do this after we have released the performance improvements to not block them.
It should actually already work, because we have not changed the konserve protocols to the outside, so it might be just a dependency issue (?).
I think it might work but the question is if you want datahike to use the konserve snapshot?!
No, you cannot use SNAPSHOT dependencies in releases on clojars in fact. But we can do a SNAPSHOT release of Datahike itself with the SNAPSHOT dependency and then release them together once everything fits. I think this is what @kordano asked for.
waiting for https://github.com/replikativ/konserve/issues/31
@TimoKramer You mean you wait for the konserve version, not the garbage collector, right?
@whilo exactly. I thought this is more or less the same.
@kordano and I agreed to use the current konserve snapshot and test all the backends (this JDBC backend, the LevelDB and filestore backend) and if it works, we can release the new konserve version.
So feel free to test the new JDBC code from Datahike and let me know how it goes.
https://github.com/replikativ/konserve-leveldb/issues/4 blocks the usage of konserve 0.6.0-SNAPSHOT
After the PR https://github.com/replikativ/datahike-jdbc/pull/12 is merged. I will change the documentation of using SQL backends in the docs and mark datahike-postgres as superseded.
Instead of using separate backends for sql backends Datahike should have one sql backend which maps directly to the jdbc drivers.