git.opendaylight Code Review

Use soft values in ValueTypeCodec

Using strong value references in STATIC_CODECS cache leads to target
classes being strongly reachable through the codec itself -- hence
they will never be weakly reachable, hence they will not be GC'd.

Use soft values, which allows the classes to be GC'd as soon as the
JVM decides they have not been used for a while -- which is what
will happen when a class becomes almost-eligible for unloading.

JIRA: MDSAL-580
Change-Id: I7956d564c46a9a1d52ba85fbe61f3a19c507a902
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 0150ddd6c4e3c61a761dacb9161e84b3f38b748a)

Bump versions to 4.0.17-SNAPSHOT

This starts the next development iteration.

Change-Id: I0281c263b67dc730f834a3f6cf89bdcd04fb2d37
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Resolve generated type for leafrefs targetting unions

We need to be able to resolve the generated type of unions
which are targeted by leafrefs during codegen.
Introduce a path that attempts this by consulting ModuleContext
for the generated types in case we are resolving a leafref.

JIRA: MDSAL-572
Change-Id: I40448a4c9f2fdf26280e0911ebb9dcc5fc60d3f1
Signed-off-by: Tomas Cere <tomas.cere@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 8c4ab985c1723336640324789c65489e71deb4e0)

Bump yangtools to 3.0.14

Pick up lastest fixes from upstream.

Change-Id: I02c935ee7a1f8c8dd6bd6f4c82e04f75fa9038f8
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Speed up BuilderGenerator matcher

We do not need to establish FQCN, as we can compare just JavaTypeNames,
which is faster.

Change-Id: I3c387e841fbf2d450195d864a928608bab8128ad
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit d613481dfab5abd014e36fc8f77b291743c57dbf)

Generate implementedInterfaces method for DataRoot

In case we have multiple top level uses statements we also
need to generate an override of implementedInterfaces() to prevent
clashes with the return types of the extended interfaces.

JIRA: MDSAL-573
Change-Id: I07ad3f6dc5a18369f0be4bdaba03c6c17072297b
Signed-off-by: Tomas Cere <tomas.cere@pantheon.tech>
(cherry picked from commit 30794cd4999beb5a7a4902d4cf0484692cbd629b)

Bump to odlparent-5.0.9/yangtools-3.0.13

Pick up latest fixes from upstream.

Change-Id: I0f07e2c35d0cb8526144f265a5711c26620a338f
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Netty Replicator - improve the reconnection and keepalive mechanisms

In some cases during a network partition the disconnected
channel got closed with delay after a new channel was already created.
This started reconnection process which closed the new channel and
created yet another one.

Also improve the keepalive mechanism since some types of network
partitions left one side unaware of the issue. It is important both
the Sink and the Source are notified about any connection issue as
soon as possible. PING-PONG messages are exchanged between the two
sides when no deltas are sent for a period of time

Configuration options keepalive-interval-seconds and
max-missed-keepalives added to both configurations.

Change-Id: Iebde72963bddb748ab97617d07cfc77cd8614da4
Signed-off-by: Tibor Král <tibor.kral@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit b94fbc60c0b41da2f6645ab51188fbcdfa74e4af)

Add documentation for Netty Replication utility

Provide more information about the replication mechanism,
instalation, configuration and usage.

Change-Id: Ifcfae6853bc1946ac9e58b84219354a543e87ed8
Signed-off-by: Tibor Král <tibor.kral@pantheon.tech>

Provide auto-reconnection for Sink

In case there is a network partition the Sink needs to
be aware of it and schedule reconnect. This patch adds
configuration knob keepalive-interval-seconds to tweak
how soon after network failure the Sink gets notified.

Change-Id: I1eb2880bb00d1101cd587e4a737ba2f8a485b7ed
Signed-off-by: Tibor Král <tibor.kral@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 2cbcb2d8589805bcb053462176424dae2b53cd1a)

Improve tests and blueprints of Netty Replication Utility

- Test replication of actual DataTreeChanges from Source
to Sink and fix any related issues.
- Move blueprints to /OSGI-INF/blueprint/ and fix any
related wiring issues

Change-Id: I89c8228538e0462bbe61ca49e1ecf09dad4d4aaf
Signed-off-by: Tibor Král <tibor.kral@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit a2f2028f4d111e77ca542d2bc2110f38fafd82fb)

Add a unit test for netty replication

This is a simple unit test, which we can use to validate establishment
of source connection. This flushes out a few bugs, which are also
addressed.

Change-Id: Ie8e5ae1772ed5dc544f434d7db44b2b31b54443b
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Remove an unused import

Since we are not running checkstyle/spotbugs, this was left out.
Make sure we enable these plugins.

Change-Id: Id043c0551c702d417be31967b27469f8b720e65e
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump versions to 4.0.16-SNAPSHOT

This starts the next development iteration.

Change-Id: Ie3b7c20b89204fd7cf404e012490becd2d14fbcf
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fixup mdsal-docs

Merge job is failing, fix it up.

Change-Id: Ifc30e344bc9f6a9548b4acb385a5fbfdb6723c4e
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 42da3dedebc639e3085da967ebec276e01e1ef2b)

Add basic netty replication utility

This adds a source/sink datastore replication component based on
Netty TCP channels. The sink (consumer) connects to source (producer)
via a channel, specifies which data tree it wants replicated and
then listens for DataTreeCandidates to arrive from the source.

Change-Id: Ib283baa9a186ae2fb4ccf909b257006d4645de37
Signed-off-by: Tibor Král <tibor.kral@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit d7b666857c54c07d9bf5c8e5e38671151d89fb4c)

Bump to odlparent-5.0.8/yangtools-3.0.11

Pick up latest fixes and improvements.

Change-Id: I54ba1e38a033afd842124573a18dbfd2a9d8a48e
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fix leafref-to-enum encoding

When we have a leafref pointing to an enum via an absolute path,
in and RPC, we end up using a no-op codec, whereas we should be
extracting the name (from enum constant, or SchemaContext).

This ends up being a problem not in codec itself, as it is using
reflection to acquire the return type, defaulting to no-op.

The problem lies with AbstractTypeGenerator, which ends up not
being able to look up the generated type in case of RPC. The problem
here is the order in which types are generated, as the leaf being
referenced is generated after the RPC itself -- hence we cannot
find the definition.

While this is again highlighting the problem of single-pass code
generation, we can side-step the problem by moving RPC/notification
generation after the root data generation.

As a further complication, we need to record the enum/schema mapping,
as the codec needs to be able to find them via leafref types.

JIRA: MDSAL-552
Change-Id: Ifd92807029cdb7ba92dcad5655bb34d3ff4cef9d
Signed-off-by: Peter Valka <Peter.Valka@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit f9eceb5081877db3fc3f840c44e943605f3f4335)
(cherry picked from commit 4ad86c646f3e447842a46b89b8697b8276942955)

Bump xtendbeans to 1.3.2

This fixes synthetic bridge method issue.

Change-Id: I3766deed8b8ed8bab5e17dc81e5fb8711c0b62e6
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 512a5596b303d6cea04e0107a258625a1efc03fa)

Document tests

Since we have brought it a dependency on test-jar, we need to
pull it to docs, too.

JIRA: MDSAL-556
Change-Id: I4914de8c8dc3cc437ec1f162352374abc99186cc
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 9b5402ab9c6fbab19ef86fa80fc2ac881b264bd2)

Cleanup DataBrokerTestModule error reporting

Do not declare RuntimeException as thrown and do not wrap reported
RuntimeExceptions. Also make sure we use IllegalStateException as
the wrapper.

JIRA: MDSAL-556
Change-Id: Ifc47004c6d5cc2c974045c3f4d94154b506c8ebd
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 4e78a1e85ca3bd04a204bb05ab3fa1695cbbf94b)

DataBrokerTestModule getBindingToNormalizedNodeCodec() & Co.

need these exposed for a new standalone component style test I'm on

JIRA: MDSAL-556
Change-Id: I5efd3a6e507d46d035067a561aaca7c2f80f51e6
Signed-off-by: Michael Vorburger <vorburger@redhat.com>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 8a6e5f0946f5ed60f7554c51e7621e2ac3e41edd)

Bug 8163: getDataTreeChangeListenerExecutor() & DataBrokerTestModule

Adjust AbstractDataBrokerTestCustomizer with a
getDataTreeChangeListenerExecutor() instead of a
setDataTreeChangeListenerExecutor(), just for more consistency with the
existing getCommitCoordinatorExecutor() method. Also less confusing (to
me) than seeing the private Executor set by default which may get
changed by the setter later.

Adjust ConcurrentDataBrokerTestCustomizer with the
useMTDataTreeChangeListenerExecutor as a constructor argument, instead
of an useMTDataTreeChangeListenerExecutor() method, just for consistency
for how you already have it in AbstractConcurrentDataBrokerTest.

Extend ConstantSchemaAbstractDataBrokerTest and DataBrokerTestModule to
allow passing through this new opt-in tweak flag, so that tests in
downstream projects such as genius and netvirt can staring exploring
enabling this.

JIRA: MDSAL-556
Change-Id: I4ad85ac48163d2f4bac865f46a3b047d5b7d333a
Signed-off-by: Michael Vorburger <vorburger@redhat.com>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 2224375d3c22043958c31195044b4169cb1caff7)

AbstractConcurrentDataBrokerTest @deprecate-s the AbstractDataBrokerTest

This is take #2 on a similar earlier attempt which made the same change
directly in AbstractDataBrokerTest, which broke some downstrean tests,
which were BADLY WRITTEN (!) because the assumed single threaded direct
execution (which DataBroker does not guarantee).

https://bugs.opendaylight.org/show_bug.cgi?id=7538 has full background

JIRA: MDSAL-556
Change-Id: I05ac3525bdcf1ab9c99dfa15b98e090848d0fddc
Signed-off-by: Michael Vorburger <vorburger@redhat.com>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 3282f1989aba438817739bbf611fd133c381a5e7)

AbstractDataBrokerTestTest which actually does something

Incl. verification that each @Test cleans up after itself. We had a
suspicion that this didn't work; it turns out that it actually does and
we had another problem somewhere else. Contributing this test still
seems useful for future non-regression.

JIRA: MDSAL-556
Change-Id: I295ebde45a0c030ab74852788b45203cdf6639e0
Signed-off-by: Michael Vorburger <vorburger@redhat.com>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 4a461d624f03984047b593c823112feac729aceb)

DataBrokerTestModule: use AbstractDataBrokerTest without inheritance

While starting to write JUnit tests (for functional components) with the
AbstractDataBrokerTest, I found myself not liking its design which
forces one to use a class SomeTest extends AbstractDataBrokerTest...

Modern JUnit tests should be written by composition, not enforced
inheritance. Thus this new DataBrokerTestModule allows one to obtain a
DataBroker suitable for Tests from anywhere.

The *(Test)Module naming convention is obviously inspired by a DI point
of view (think Spring Framework's @Configuration classes, or Google
Guice or Dagger (yay!) *Module classes - which is exactly what this
really is - a particular way to obtain an instance of an
implementationof some service (here a DataBroker) in a certain
environment (here for tests).

The internal implementation of DataBrokerTestModule could be changed
later; I guess ideally what's in AbstractDataBrokerTest could go into
DataBrokerTestModule and AbstractDataBrokerTest could use it, but for
now I'm too lazy to change that, as this does the trick nicely. Later
implementation changes would be transparent to users of
DataBrokerTestModule.

JIRA: MDSAL-556
Change-Id: I9641f527bbc0cb92732f2e513cdd64cc6a837200
Signed-off-by: Michael Vorburger <vorburger@redhat.com>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit e94d340da0d1b19e506fe2ad7e4e19c79959e241)

Fix SnapshotBackedWriteTransaction error message

The message is misleading, as we are not performing a write, but
a merge operation.

Change-Id: I8b61ac7b0bf5f3ae746a2b6c1883a02e6fe61e47
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit fb0a271c02454f11cd3c64130513dfb7caf3f386)

Bump versions to 4.0.15-SNAPSHOT

This starts the next development iteration.

Change-Id: Ie328cd001fb3bf9f45f174ef6c22fd6765fd7802
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump to odlparent-5.0.7/yangtools-3.0.11

Adopt latest releases of upstream in preparation for next release.

Change-Id: I3c57861bc567da22240678bb7877f071d0cc8123
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Add test-sources for mdsal-binding-dom-adapter

We are (unwisely) constructing a test-jar here, hence we should
also be publishing the corresponding sources.

Change-Id: I08ffb23b1e5eb8036e16237b8bc6fb0c6480f848
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 9cf191a88781e5208392b08740e480fd048e58c2)

Improve notification delivery defensiveness

If a listener throws an exception, make sure to log it and proceed
as if nothing happened.

JIRA: MDSAL-544
Change-Id: I0ca54bb5c517115b840ad03ea24d442c7762ff05
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit bc58fd6b46acb6f94d6b9395a67f7112d134f0c1)

Fix unsubscribe checks in DOMNotificationRouterEvent

The checks here are based on old assumption that the listener
would be set to null, which does not hold true for quite some
time. Update the check to close a simple race.

JIRA: MDSAL-545
Change-Id: I950ffcbcf732d4c24dcad675b81c2889465d9045
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 91749a5a5fb089e74306f288d786acb8d3c450ae)

Improve grouping sort allocation

Pre-allocating the result list saves us the common case when
the input is actually empty.

Change-Id: I237a75db38e8e0c160b204875bc5824433e9c3dc
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Improve grouping error reporting

Reporting FQCN here is useful, as it is giving us crucial information
as to what class is being generated -- including the namespace.

Change-Id: I1ad5f6a0a568a71656e338ae0d5d9e3f863103a4
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Cache YangModuleInfo type name

Analysis of a heap dump shows we are creating a number of instances
pointing to the same type for the purposes of having the QNAME constant.

As we have the context in which we are creating YangModuleInfo,
instantiate a primitive cache to de-duplicate these definitions.

JIRA: MDSAL-515
Change-Id: I9bb52d04627e91eac3adff672b06f6340cac157f
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fix logic ensuring parent presence

When we are asked to create intermediate nodes for a path, whose
parent is the conceptual root (i.e. its parent is empty), we should
not attempt to merge it, as it just does not make sense.

JIRA: MDSAL-534
Change-Id: Ia5f6ef2857903d8e88030f4c1f3867b8ed8f5a0e
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit ed271c6322092bccf8ff9500b1bbb0fb62683419)

Remove superfluous @NonNull annotations

These annotations are inherited from interface contract, there is
no point in specifying them again.

Change-Id: I7e2cb6d353b358f5c2eefc03181803377e047750
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 9918388ee26503d8516e3e6d81ac2e82794f6322)

Migrate test to use mergeParentStructure{Merge,Put}

We have deprecated these methods, make sure we migrate the test
so that it does not add false warnings.

Change-Id: I3a6bd6b936f34478fe7afdcc26547ffa7edb8992
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 77998716f93d97aa8e58e408db6d956d81b727ff)

Bump versions to 4.0.14-SNAPSHOT

This starts the next development development iteration.

Change-Id: I5289dc3331cc03eb66676e732b845d4fe351abff
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fix yangtools mis-bump

I7e374903ca22059f5e31377a9b91971a2d82e196 actually failed to pick
up yangtools-3.0.10, fix that up.

Change-Id: Iece7d3a7bf15db6e3f287e86bbb35f9f8749482b
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump versions to 4.0.13-SNAPSHOT

This starts the next development development iteration.

Change-Id: If3b2159f6e106064d0deee1acb13bf30bdc93d6a
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fix YANG snippet escaping

This is the non-javadoc path for module generation, where we are
failing to escape potential unicode references. Add the proper
callout and also optimize it to be a faster no-op when not
applicable, properly documenting what is going on.

JIRA: MDSAL-529
Change-Id: Ib21ed4f15555d49aaa7eb3dac61043c4b2e90451
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump to odlparent-5.0.6/yangtools-3.0.10

This picks up latest fixes from both upstreams.

Change-Id: I7e374903ca22059f5e31377a9b91971a2d82e196
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Propagate grouping inference flag

The test case ends up looking through the magic code flagged
for refactoring -- which is losing knowledge that the parent
node is actually a grouping. Fix this up by propagating the
flag correctly.

JIRA: MDSAL-519
Change-Id: I8fe04e688adfcc068ff29e90a039530bcd56b3c3
Signed-off-by: Samuel Kontris <samuel.kontris@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Expand test suite to cover uses-augment of actions

Augment can also target a relative path from uses context, this
expands the testing model to cover that case, too.

The test case flushes out the problem of not handling
action/notification nodes when targeted in this manner, leading
to an obvious hiccup.

JIRA: MDSAL-517
Change-Id: If389440622c31f544917b04e3825fe6e08e08244
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Update Mdsal517Test a bit

This just cleans up naming to prevent confusion.

JIRA: MDSAL-617
Change-Id: I75d2e95d2396f63d05cd533f16057f556850763c
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fix augmentation pointing to a grouping action's input

When augmentation targets an action input which itself is defined
in a grouping, we need to actually lookup the original definition.

Add a type aliasing information, which should be consulted when
making a simple reference to a type -- and use it in
augmentationToGenTypes().

JIRA: MDSAL-517
Change-Id: I1ef383e8777a7c09a979cbdd84f71be3a0f9b021
Signed-off-by: miroslav.kovac <miroslav.kovac@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Switch from NPE to ISE

When we fail to locate a type it constitutes an IllegalState, not
a plain NPE.

Change-Id: I46eaa470342a5e72ad6a92aefed97e458a2ca398
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fix action/grouping lookup mechanics

When we are following references towards the original definitions
we need to mind the fact that 'uses' statement may point to a grouping
in a different module and hence we need to adjust namespace accordingly.

The same thing applies when we want to find generated types for such
actions' input/output -- here we need to acquire the correct
ModuleContext.

JIRA: MDSAL-516
Change-Id: I1931279beb8e663e6d8558d50c5d7d274e248cf3
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Use osgi-core constants

Just to keep things consistent, use OSGi's Constants class to get
the bundle id for framework (i.e. SYSTEM_BUNDLE_ID).

Change-Id: I18dae474c944f29ebb32203a12db1c96c2e1483f
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit a246443e10cc43adace7bc36fdf8014d76010389)

Deprecate BindingRuntimeContext.getEnumMapping()

Generated enumerations capture their YANG-declared name and expose
it as their method.

JIRA: MDSAL-392
Change-Id: Iac16b6b3599565f059332ed079c432c0603ac500
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Refactor GeneratedClassLoadingStrategy

This class holds two implementations, each of which is a singleton.
Make sure we lazy-load implementation classes so that the CHA can
be more effective.

JIRA: MDSAL-392
Change-Id: Ia31f63d877ec1025e1a06bf8676d918857e6fd1e
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fix checkstyle a bit

We have a misaligned method here, fix that up.

Change-Id: Iff4ad3e9a5c2da1e64f73f8b3dfaf78d77b1db85
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump versions to 4.0.12-SNAPSHOT

This starts the next development iteration.

Change-Id: I9a389d9c99840de5d0cf8766793a5f2bbaab7836
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Fix path namespace resolution

If a path is defined in terms of a submodule-local import,
SchemaContextUtil.findDataSchemaNode() breaks down because
it's reparse attempt fails to find the import in the actual
module.

SchemaContextUtil is providing findDataTreeSchemaNode() which
works properly on parsed PathExpression, use that instead.

JIRA: MDSAL-499
Change-Id: Ifa8a8958d7118d851d27fcc2c70c6d8382e3e3c0
Signed-off-by: miroslav.kovac <miroslav.kovac@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump versions to 4.0.11-SNAPSHOT

This starts the next development iteration.

Change-Id: I4017bdc625b2e70308995f023c4e8e569a96d4a8
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump to odlparent-5.0.5 and yangtools-3.0.9

Pick up latest updates and fix for SFT.

Change-Id: Ie547285ce14a943107646585d9a24666f6433025
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump versions to 4.0.10-SNAPSHOT

This starts the next development iteration.

Change-Id: Ic1b4b1949741087893ae3f23e62a320f07a9231f
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Migrate away from DataNodeIterator

We do not need to collect the entire analysis, just typedefs
(and containers in a test). Migrate to new SchemaUtils methods which
provide exactly that functionality more efficiently.

Change-Id: Idfbe19c785351f3280eae33c476d533055cb16e6
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump yangtools to 3.0.8

Pick up latest fixes and improvements.

Change-Id: I9e7e6d236c8059ca42220117aa49c88c7f894b79
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Use parallelStream() to generate files

This is a follow-up cleanup, reducing complexity of parallel dispatch
by taking advantage of parallel streams.

Change-Id: I7f5edd1444f171fba9b7118531678875a8692223
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit c0c76a53d25048d86fcd60e16b68fe942672b85a)

Generate binding files in parallel

Binding-generated files are standalone in that the process of
generating its source does not have any other side effects.

Improve the performance by generating these files in parallel
using the common ForkJoin pool -- thus taking advantage of all
available cores.

Change-Id: I4ee9efea0371fe9574721701ea9674793209cf24
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit a6fb0b3903e8630df18b7c63a116bfd06a716bdf)

Add utility methods to strip zone from Ipv{4,6}Address

Most callers really want to operate of Ipv4AddressNoZone, as a zone
does not make sense. Some amount of translation may be needed when
dealing with models which we have no control over, hence ensuring
no zone is present needs to be done quickly.

This adds ipv4AddressNoZoneFor(Ipv4Address), which does an efficient
check (for the case where the source is already a NoZone) and strips
the zone.

Change-Id: Ieada4625637d0216983b515324bb5d33f5d33a99
JIRA: MDSAL-510
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 6c02ae5ce070dfeeeaae878dc5436c601d305886)

Parse IPv4-mapped IPv6 address as an IPv6 address

Inet6Address.getByAddress(byte[]) does not explicitly exist, and is
really InetAddress.getByAddress(byte[]). That method ends up returning
an Inet4Address for IPv4-mapped IPv6 addresses.

This in turn causes addressStringV6() to do the wrong thing and format
the address as an Ipv4. This is not caught by validation, as we are
skipping it by default -- and ends up causing problems way later when
such an address meets an enforcement point -- such as DOM->Binding
translation.

Fix this by using Inet6Address.getByAddress(String. byte[], NetworkInterface),
which is guaranteed to result in an Inet6Address.

Also add proper defences to addressStringV6(InetAddress), so that any
attempt to mis-use it is caught.

JIRA: BGPCEP-891
Signed-off-by: Ajay Lele <ajayslele@gmail.com>
Change-Id: Ife4cc9e57bd49b45f2beb73462a865ad84483d8f
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 1d8897caf5cec5f044f870f949a15fa57de54d71)

Eliminate shift variable

We do not need to track shift amount, as we can can just shift
the entire previous result, leading to more invariants in the loop.

JIRA: MDSAL-509
Change-Id: I13a108be27cac2fe4498952f3f05437927c999c0
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 4b26656c945b9af5e67d28d10b014b0ed37cc510)

Add DottedQuad/int interface

DottedQuad is essentially an Ipv4Address, hence we should have
the same level of support. Add interface to create DottedQuad from
bits and serialize it to bits.

JIRA: MDSAL-509
Change-Id: I9cee4afa1993a0090f10b3969152344e03e9adc0
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 4e1311cbb83dd15f2da7936385dd9384b6dc0229)

Add IPv4 address bridges to primitive int

Interfacing Ipv4Address(NoZone) with plain int is useful when
we we want to interface with byte streams, such as DataInput
or Netty's ByteBuf. This adds the methods to do so.

JIRA: MDSAL-509
Change-Id: Icce95e4a94ae41be4fc2aee352f83655ea55e89f
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 98143cfc18f95eb4057587c1bdb4aa3b5e3b2e20)

Add HexString/DottedQuad/Uuid support to IetfYangUtil

HexString is similar to a MacAddress and PhysAddress, hence it is
very simple to support it.

DottedQuad is essentially an Ipv4Address, hence it is almost as
simple to support it, too.

Uuid can be created from java.util.UUID, so that is the input
we are expecting.

JIRA: MDSAL-508
Change-Id: I82abd68b23bf857deeb9a64eefb69164fe6c8b24
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 7cbd12d8894b1703b19e70b07436a937a7b8d0bf)

Cleanup AbstractTypeProvider.isLeafRefSelfReference()

This method's logic eagerly strips the xpath, even when we may
end up not needing it. Refactor implementation to initialize
later (and bail quicker).

Change-Id: I7a54d14dcc3def1cc01f65523c5cf6e1277a0fce
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit c8c1ed80b2949b5eda8bdb57bc9a25ef0db1aef0)

DTOs for anydata/anyxml are not generated within a list

Code generation for container and list statements follows different
code paths and we have missed a hook in the latter to generate
opaque objects.

JIRA: MDSAL-506
Change-Id: I25a54362ceb09937ffe796081660afe7fa1d6bb2
Signed-off-by: wusandi <wusandi@163.com>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit e905b3d4792076251ddc276fa248e441bba157b2)

Speed up AbstractBuilderTemplate.removeProperty()

Rather than performing a linear unbounded search followed by a
conditional remove, use an interator and remove the first matching
property, bailing out.

Change-Id: Ia5de2175628c9730d7c76df58e15540f3cbb4fa3
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit a5f323b5a0eeb5d4a0cb82b37f751ae780ba1f91)

Use BindingTypes.identifiable()

Instead of looking up Identifiable type each time we use it, use
the convenience method exposed through BindingTypes, improving
efficiency a bit.

Change-Id: Ib8e7bf0385a51ee687484a2667905e284a90ab3a
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 06c6f6b581b0d1526231d84cf0884142fc4f8a4c)

Add BindingTypes.augmentation()

This migrates Types.augmentationTypeFor() into BindingTypes, so we
have proper separation.

Change-Id: I6ff923be4b598714761ed60ca6a36037295b9ac5
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 2f8f56af98be1f58d14256e780e2a9a77c0cfc49)

Deprecate Types.augmentableTypeFor()

This method has a mirror in BindingTypes, which is a more appropriate
place for the definition. Deprecate the method for removal and
redirect it to BindingTypes.augmentation().

Change-Id: I462a2fe25f0b53cca0619c72aac7a1e602718143
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit a07668d09b87d8f671e774a13583a3c3c0b41369)

Add BindingTypes.QNAME

We are looking this type for each constant, which is not entirely
efficient. Add a constant to BindingTypes and use it directly, providing
a slight performance benefit.

Change-Id: I56c33750b085728a80782da93e9ea4a250b988aa
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit d9896900305375bebef22f93e36b713408062fd5)

Fix RPC method name conflict with JLS

When we are constructing the method name for use with the RPC
provider interface we need to pay attention to conflicts with
JLS reserved words.

If such a conflict occurs, append a single '$' to the name, so
that we have a valid method name.

JIRA: MDSAL-500
Change-Id: Ifb533e2daa71bc9d1780a5d8eb48da6ce3b2d209
Signed-off-by: miroslav.kovac <miroslav.kovac@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 6e6ebc43866204667658f8bbf7cfb6233caffdc7)

Bump versions to 3.0.9-SNAPSHOT

This starts the next development iteration.

Change-Id: Ic3476c2c1b8e06f8034790a9b2c695504e53ff51
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Bump odlparent/yangtools to 5.0.4/3.0.7

This picks up the latest upgrades.

Change-Id: Ib02e30c7dcfa4ded9fa3b2e13015cde0b271b2c9
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Add codegen support for nested notifications

We do not have a concept of instance notification, which is similar
in many respects to an action. Add the baseline interface marker
and teach mdsal-binding-generator-impl to emit interface definitions
for them.

JIRA: MDSAL-493
Change-Id: I6af84b50f3083a8e1046ee0521b126aae1ffafe5
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 480afab4e0a0eb8d64367c0bd12fa3ec892a3383)

Correct @NonNull CodeHelpers.compilePatterns()

Arrays and type-use annotations have weird way of being written,
this adds the correct form.

JIRA: MDSAL-487
Change-Id: I2e31c90eca5b59250f0a9e33444fa8b3b8ae5e4c
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit c03c60cc068b342582a2d54861719718c750a880)

Fix CodeHelper.nonnull() nullness annotation

We do not correctly propagate the fact the return is @NonNull,
leading to Eclipse issuing warnings. Fix that.

JIRA: MDSAL-487
Change-Id: I24360192d105423a19165855521bb24bb0bd0678
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 8fc4a97304080fafab158ef877e6fc180c030ac0)

Correct YangModuleInfo.getInstance() nullness warning

This method is missing a @NonNull, which means we end up issuing
a warning. Fix that up.

JIRA: MDSAL-487
Change-Id: I0870880bceb727b6a820bcdc0fd4177002fb4119
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 1144f464d67c5371b30463be5e4e144edaddd05b)

Fix onDataTreeChanged() references

Use a link to specific method, not just any matching.

Change-Id: I99752bbda0e9ff3d3371ba8658b319a9cbfe4fce
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit f23c00c51d7874475c91a9290b561a9ab1d0cdc5)

Add yangtools javadoc link

Cross-references mdsal-docs with yangtools-docs.

Change-Id: Iea811968568f48de9070a0a45876db71dce37470
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 674ce26e4f2c3e7f7400f7920b78bef657c05a0a)

Resync javadoc links

This re-synchronizes links to upstreams to match odlparent.

Change-Id: Ibc514a9aa5eb5f068561772b91061567796ed171
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 24dc1f6ea4ccdf9f8360139c23fcc3988aeabd26)

Remove use of xbase-lib in tests

The way the expected object is constructed is horrible, just use
plain fluent builder without the extensions. This makes it more
clear and much more concise.

Change-Id: I042dfda9fadcc2b3764a449dfc34300f0928370b
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>

Import types being checked

When we are checking individual component types, we should not
be needing to reference them via the full name. Use imported name
instead.

Change-Id: I728d4790929cb548ef022c3b3b34e6cddc2f5aa3
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit b76fcb3bb19ecffdfbb47d7840a51e335bead392)

Add Types.isListType(ParameterizedType)

Performing duplicate checks is not efficient, if the caller already
knows the type is parameterized.

Change-Id: I164291490602959d1f5bdd83eb35be7f2b0d0428
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit c7da6941c5e7f46e189cc85a8b4eb66ba8551f70)

Speed up ListsBindingUtils

This just uses proper fluent builder pattern instead of a
temporary variable.

Change-Id: I8cdf2cff7c910f5518eb999949f315740fe7c1ca
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 9e4b1ee522aa173efe114e315b017f77716cdb79)

Inline AbstractTypeGenerator.resolveListKeyTOBuilder()

This method is only called once in a context which can already
know whether it needs to actually call it. Inline the definition,
so that we have a more efficient code flow.

Change-Id: I216ca272a13d98eb3ce4c6323fe9442c9ecaa617
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 60a335a9fcbf4ee514d32981c4e03113b115dc1e)

Optimize AbstractTypeGenerator.listKeys()

In case of an empty key definition we end up allocating
short-lived empty ArrayLists, which is not efficient. Also lists
typically have a few elements, so we end up over-allocating them.

This fixes both cases, by checking the definition first and
returning an empty list in case the definition is empty, as well
as properly allocating efficient implementation where needed.

Change-Id: I99e8843abf6040654fab96052352cbfb29590ecf
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit efe9042ecb51521850d9b90f4d8460a35c9bd9ea)

Remove unneeded checkstyle supressions

Treading carefully and using Class.asSubclass() allows us to
remove raw types/unchecked casts.

Change-Id: Ib7554cc40f0b4f41495405db3b7715e0f4e21cc7
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 6087dda3aa2e148c5046fb2d8e7b0f75a5eee844)

Parse docs version of pom.xml

Issue: MDSAL-489
Signed-off-by: Thanh Ha <zxiiro@gmail.com>
Change-Id: Ie32ef32b3d89e98b9ea432bd8e2ac5ea91bf13b9
(cherry picked from commit da3cf4cd9f78d66a7edac29d6efa74a7b8a867b6)

Specialize JavaFileTemplate.importedName(Type)

95% of call sites do not use annotations, hence we specialize
this method and support it through specialized
AbstractJavaGeneratedType.getReferenceString(Type).

This clearly separates the two codepaths and removes superfluous
checks and indirections in both paths.

Change-Id: I17bfef631d446d71147b17146d5b2cf71a8be623
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit dbaae85acd37ba622f4f53a7335af14e768fb180)

Use SUPPRESS_WARNINGS reference

When suppressing warnings, we should use the JavaTypeName constant
available in JavaFileTemplate.

Change-Id: I9b6a46403b67f4411356e2aaf16c758b15430273
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit b5a4b1eb66e208a957cef4ff43fe58a68af4c928)

Add @NonNull annotation to $YangModuleInfoImpl.qnameOf()

YangModuleInfoTemplate has a lot of ceremony when we statically
know the types being used by the template. Concentrate imports
into two sets, instantiate them as strings and be done with it.

Since this takes care of most dynamics, making things much easier
and predictable, we can easily add @NonNull annotations as needed.

JIRA: MDSAL-487
Change-Id: Ic9cd4da1002f3edea6f3fb18f9d49e4a377e145b
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit b0ea6f0f70fbc597779703a4c65dd85b69355de3)

Cleanup YangModuleInfoTemplate

generateModelProvider() returns a string, use the appropriate
template syntax. Also use Map.putIfAbsent() instead of
containsKey()/put() checks.

Change-Id: I48faccd7c87f2ffff51b30d9df04c5cb613df184
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit d8bc8182155d7ee25ab056c487a4e09b5b2e8df6)

Fix spotbugs complaints around logging

Spotbugs is very much correct about warning about non-constant
format strings. This patch fixes that up.

Change-Id: I8df61b64be1478510e60af99dd2cd2365a98f2c1
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
(cherry picked from commit 89ffe8626f4addebedc19f4c85e12d35d047b2e0)