Closed jeffdyke closed 5 years ago
What would make you comfortable merging this? Or are you now?
Hello @takezoe - A couple things: 1) I'm not very familiar with Mochito so am not having a good time fixing the tests, i'd love some assistance, so then i could write more tests.
2) Would you be willing to publish this as a snapshot if not direct release. I have some ideas about how to make this safer, but don't want to start until we agree on this.
I have published this to my own github b/c i required this functionality, at least to be safer and raise more errors when they occur and show results rather than swallowing them with unit
. I have left it under your namespace, so we can continue to share. Somethings will conflict, but i needed to move forward.
@jeffdyke Sorry for my late reply. I'll try to fix failing tests.
@jeffdyke I created #55 to fix failed tests (and added some new tests) without last 3 commits of this pull request. Is it good for you? If so, I'm going to merge #55 instead of this, and release the next version (or SNAPSHOT).
@takezoe first thank you. Secondly yes #55 makes more sense as i should have re-branched to self publish.
OK. I merged #55 and closed this pull request. I'm going to release 0.0.20.
Released.
Per https://github.com/takezoe/solr-scala-client/issues/53, i have added a
collection
method for all calls as well as tidied upBatchRegister
to more cleanly call the appropriate method. I found that when sending a null in (i swapped that out forNone
), it would still use the method that takes a collection, so the NPE still exists. I could be wrong about that portion, but the match felt the safest to me.I think it's really unfortunate that the JavaClient allows for both instantiations as it leads to incredibly unsafe code, especially since the client is so cheap to create. Personally i'm now creating the client with no collection in the URI and using all the
collection
methods.I've tested this thoroughly and will add unit tests if this approach is is accepted.