You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
syndtr/goleveldb doesn't appear to be maintained anymore. There's a compaction timer leak that was fixed in ucwong/goleveldb, and there appears to be a variety of PRs opened in the syndtr/goleveldb repo that provide some useful optimizations.
Really the decision is what fork do we use? Should we use our own fork that we can merge improvements into?
LevelDB is the key-value store used by the reference counter, so we need to make sure that any fork we switch to doesn't introduce any unwanted side effects, or break compatibility. I think there's a case to be made for us forking goleveldb, and maintaining it ourselves since we use it so heavily, however I would rather to keep our forked repositories as minimal as possible unless there's specific cases to be made for maintaining our own, which this somewhat seems to have.
I prefer to use other people's work as much as possible. We should use ucwong's fork. If our needs deviate, we can always switch to our own fork.
Checking on the diffs, the fork should be fully compatible, and it's easy to guaranty with go-datastores' test-suit. ucwong is also a long time active developer on Github.
syndtr/goleveldb
doesn't appear to be maintained anymore. There's a compaction timer leak that was fixed inucwong/goleveldb
, and there appears to be a variety of PRs opened in thesyndtr/goleveldb
repo that provide some useful optimizations.Really the decision is what fork do we use? Should we use our own fork that we can merge improvements into?
LevelDB is the key-value store used by the reference counter, so we need to make sure that any fork we switch to doesn't introduce any unwanted side effects, or break compatibility. I think there's a case to be made for us forking goleveldb, and maintaining it ourselves since we use it so heavily, however I would rather to keep our forked repositories as minimal as possible unless there's specific cases to be made for maintaining our own, which this somewhat seems to have.
@xiegeo thoughts?
The text was updated successfully, but these errors were encountered: