[jira] [Commented] (IGNITE-2560) Support injections in entry processors
[
https://issues.apache.org/jira/browse/IGNITE-2560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15432965#comment-15432965
]
Andrew Mashenkov commented on IGNITE-2560:
--
Hi, Semen
{quote}
I can't find test for invokeAll(Map> map, Object... args)', could you please point me where it was added?
{quote}
See class _GridCacheAbstractFullApiSelfTest_. Method
_testTransformResourceInjection_ calls _doTransformResourceInjection_ with
cache and cache-with-async. Method _doTransformResourceInjection_ make checks
with _invoke_, _invokeAll_, _invokeAll (Map ...)_.
{quote}
in GridLocalAtomicCache let's do not use wrapper at all, instead just call
GridResourceProcessor.inject on entry processor
{quote}
Please, take a look on method _invokeAllAsync_ in GridLocalAtomicCache class.
Method signature tells it throws *no* exceptions, but
_resourceProcessor.inject(..)_ does. Whether exception is being processed
correctly?
> Support injections in entry processors
> --
>
> Key: IGNITE-2560
> URL: https://issues.apache.org/jira/browse/IGNITE-2560
> Project: Ignite
> Issue Type: Improvement
> Components: cache
>Reporter: Valentin Kulichenko
>Assignee: Andrew Mashenkov
> Fix For: 1.8
>
>
> Currently resources are not injected in entry processor, which is not
> consistent with other functionality, like closures, jobs, listeners, etc.
> To avoid performance degradation we should introspect the class only once,
> cache this information and do not try to inject if there are no annotations.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (IGNITE-2560) Support injections in entry processors
[ https://issues.apache.org/jira/browse/IGNITE-2560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15426757#comment-15426757 ] Andrew Mashenkov commented on IGNITE-2560: -- Hi, Semen PR has been merged from latest master. Code's been cleaned up. I didn't run any benchmarks. I saw benchmark tests implemented by Alexey Stelmak, but I have no idea how to run it. Any help would be appreciated. > Support injections in entry processors > -- > > Key: IGNITE-2560 > URL: https://issues.apache.org/jira/browse/IGNITE-2560 > Project: Ignite > Issue Type: Improvement > Components: cache >Reporter: Valentin Kulichenko >Assignee: Andrew Mashenkov > Fix For: 1.8 > > > Currently resources are not injected in entry processor, which is not > consistent with other functionality, like closures, jobs, listeners, etc. > To avoid performance degradation we should introspect the class only once, > cache this information and do not try to inject if there are no annotations. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (IGNITE-2560) Support injections in entry processors
[ https://issues.apache.org/jira/browse/IGNITE-2560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15411764#comment-15411764 ] Alexey Goncharuk commented on IGNITE-2560: -- The pull request is https://github.com/apache/ignite/pull/510 > Support injections in entry processors > -- > > Key: IGNITE-2560 > URL: https://issues.apache.org/jira/browse/IGNITE-2560 > Project: Ignite > Issue Type: Improvement > Components: cache >Reporter: Valentin Kulichenko >Assignee: Alexey Stelmak > Fix For: 1.7 > > > Currently resources are not injected in entry processor, which is not > consistent with other functionality, like closures, jobs, listeners, etc. > To avoid performance degradation we should introspect the class only once, > cache this information and do not try to inject if there are no annotations. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (IGNITE-2560) Support injections in entry processors
[
https://issues.apache.org/jira/browse/IGNITE-2560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15166318#comment-15166318
]
Alexey Stelmak commented on IGNITE-2560:
- ??for me all new tests fail with error??... fixed.
- ??can not find changes for local/atomic cache?? done.
- ??I think it makes sense to add flag 'entryProcessorAnnotations' in
GridResourceIoc.ClassDescriptor to avoid extra checks on each 'invoke' call
(and remove method 'isAnnotationsPresent' you added)??... done, but implemented
differently (see 'isAnnotationsPresent')
- ??need to add tests for both methods 'invokeAll(Map map)' and 'invokeAll(Set
keys, EntryProcessor entryProcessor)' and check in tests all supported
annotations??... done (GridCacheAbstractFullApiSelfTest and
GridTransformSpringInjectionSelfTest)
- ??need test all transaction types and not only case when tx is rolled back,
also for tx cache need check case without starting explicit transaction (just
'cache.invoke' outside of IgniteTransactions.start()/commit)??... done.
- ??please change exception thrown by EntryProcessorResourceInjectorProxy:
'throw new RuntimeException(e)' -> 'throw new IgniteException("Failed to inject
entry processor resources', e);'??... done.
- ??inside GridCacheMapEntry.innerGet0 and innerUpdate entry processor class
name is passed to EVT_CACHE_OBJECT_READ, when proxy is used then original entry
processor class name is lost, need to fix this and add test??... done.
> Support injections in entry processors
> --
>
> Key: IGNITE-2560
> URL: https://issues.apache.org/jira/browse/IGNITE-2560
> Project: Ignite
> Issue Type: Improvement
> Components: cache
>Reporter: Valentin Kulichenko
>Assignee: Alexey Stelmak
> Fix For: 1.6
>
>
> Currently resources are not injected in entry processor, which is not
> consistent with other functionality, like closures, jobs, listeners, etc.
> To avoid performance degradation we should introspect the class only once,
> cache this information and do not try to inject if there are no annotations.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (IGNITE-2560) Support injections in entry processors
[
https://issues.apache.org/jira/browse/IGNITE-2560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15160344#comment-15160344
]
Semen Boikov commented on IGNITE-2560:
--
Reviewed, there are some problems:
- for me all new tests fail with error:
{noformat}
Caused by: java.lang.IllegalArgumentException: Unwrapping to class is not
supported: class org.apache.ignite.internal.processors.cache.GridCacheContext
at
org.apache.ignite.internal.processors.cache.CacheLazyEntry.unwrap(CacheLazyEntry.java:146)
at
org.apache.ignite.internal.processors.cache.CacheInvokeEntry.unwrap(CacheInvokeEntry.java:131)
at
org.apache.ignite.internal.processors.cache.EntryProcessorResourceInjectorProxy.process(EntryProcessorResourceInjectorProxy.java:68)
at
org.apache.ignite.internal.processors.cache.GridCacheMapEntry.innerUpdate(GridCacheMapEntry.java:2186)
{noformat}
- can not find changes for local/atomic cache
- I think it makes sense to add flag 'entryProcessorAnnotations' in
GridResourceIoc.ClassDescriptor to avoid extra checks on each 'invoke' call
(and remove method 'isAnnotationsPresent' you added)
- need to add tests for both methods 'invokeAll(Map map)' and 'invokeAll(Set
keys, EntryProcessor entryProcessor)' and check in tests all supported
annotations
- need test all transaction types and not only case when tx is rolled back,
also for tx cache need check case without starting explicit transaction (just
'cache.invoke' outside of IgniteTransactions.start()/commit)
- please change exception thrown by EntryProcessorResourceInjectorProxy: 'throw
new RuntimeException(e)' -> 'throw new IgniteException("Failed to inject entry
processor resources', e);'
> Support injections in entry processors
> --
>
> Key: IGNITE-2560
> URL: https://issues.apache.org/jira/browse/IGNITE-2560
> Project: Ignite
> Issue Type: Improvement
> Components: cache
>Reporter: Valentin Kulichenko
>Assignee: Alexey Stelmak
> Fix For: 1.6
>
>
> Currently resources are not injected in entry processor, which is not
> consistent with other functionality, like closures, jobs, listeners, etc.
> To avoid performance degradation we should introspect the class only once,
> cache this information and do not try to inject if there are no annotations.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
