* Sprinkled new gradle dust to make both IntelliJ and TC (gradle) compilation happy.
* Fix compilation problem once and for all + ignore the RPCProxyWebServiceTest smoke test.
* CORDA-1393: Install `onError()` handler for folding action
or else `ErrorNotImplementedAction` will be invoked which is never a good thing
* CORDA-1335: Improve exception handling in `cleanUpOnConnectionLoss()`
* CORDA-1335: Try to trick the logic to pretend we are running in HA mode to have a chance of re-connecting.
* CORDA-1416: Make `NodeMonitorModel` code react to proxy changing.
* CORDA-1416: Workaround `CordaRPCOps.equals()` calls when listener dispatching change.
* CORDA-1416: Increase re-try interval to allow enough time for server to come back online.
* CORDA-1355: Properly close RPC connection we are moving away from.
* CORDA-1355: Unsubscribe on Error to prevent propagation of it downstream.
* CORDA-1355: For downstream subscribers ignore errors properly. Thanka to @exfalso for the hint.
This fixes: Transaction Updates do not flow after re-connect
* CORDA-1355: Bugfix eliminate duplicating items on "Transactions" blotter after re-connect.
* CORDA-1355: Bugfix eliminate double counting on dashboards.
* CORDA-1355: Bugfix eliminate same parties in dropdowns.
* CORDA-1355: Stop using `SecureHash.randomSHA256()` for painting widget icon.
Instead use combined SHA hash such that icon represents the whole population of trades.
That way two transactions blotters can be compared by a single glimpse at corresponding icons.
Also minor refactoring.
* CORDA-1416: Make RPC re-connection faster/more robust.
* CORDA-1416: Properly announce thet Proxy may not be available during re-connect and prevent UI crashing.
* CORDA-1416: Disable UI until RPC proxy is available.
* CORDA-1416: Correct typo.
* CORDA-1416: Unit test fix.
* CORDA-1416: GUI cosmetic changes.
* CORDA-1416: Correct spaces.
* CORDA-1416: Remove un-necessary overrides in CordaRPCOpsWrapper.
* CORDA-1416: Switch from using `doOnError` to installing an error handler upon subscription.
When the config file specified doesn't exist, a stacktrace is dum
ped to stdout rather than just printing a nice error message. Same
for specifying a stupid mode.
The nice fix would be if the cmdline parseing library exposed the error
states from Converters properly. Alas, what it actually does is catch
the exception you'd like and throws a very generic exception.
So, catch and rethrow with a better error message and don't dump the
stack (i.e. exit rather than rethrow)
Database transaction can be set to retry failure due to any exception with a cause or a nested cause of SQLException type.
Rationale: By the default transaction is retried only for SQLException. It may happen that SQL Exception is wrapped by Hibernate exception, allow to retry such cases if requested e.g. database.transaction(recoverableFailureTolerance = 3, recoverAnyNestedSQLException = true) { .... }
* Prepare node-api for determination.
* Disentangle Kryo and AMQP classes.
* Add version properties for fast-classpath-scanner, proton-j and snappy.
* Remove String.jvm extension function.
* Refactor Cordapp reference out of AMQP serialisers' primary constructors.
