Andy Seaborne
Andy Seaborne
Thanks for raising this issue. Jena provides: `GraphTxn` from `GraphFactory.createGraphTxn` (new name `Factory` -> `GraphFactory`). `GraphTxn` is thread-safe on each single method call even without calling transaction begin-commit/abort from the...
From the Jena project point of view, it's not desirable to add implementations that support one or two uses if there is a way to use the general machinery. The...
The trouble with locks is deadlocks - both absolute ("deadly embrace") and under load ("deadlock by congestion"). Don't mix "simple" and "lightweight" :-) Java locks are not lightweight. Their contract...
> it is difficult to me to imagine a scenario with deadlocks The discussion here is about general support. If graphs implementations are in ONT-API, they can do what is...
Side-effect of the PR. Reopened.
GraphTx does more. Because that includes multiple true concurrent views of the data it will cost more. It does not require the Kotlin runtime. This is like "autocommit" for SQL...
> I see that GraphTxn no longer works properly Is that stacktrace for the GraphTxn case? I can't see where the "find" step is. Why is the code writing inside...
> the documentation update is in progress: https://github.com/apache/jena-site/pull/182 is merged @Aklakan Is this issues finished now?
Related issues: * #2151 * #1968 * [JENA-2346](https://issues.apache.org/jira/browse/JENA-2346)