Hi!
apologies for the extra wide distribution (this exhausts my once
a year ASF mail-to-all-bigdata-projects quota ;-)) but I wanted
to suggest that all of you should consider submitting talks
to FOSDEM 2017 HPC, Bigdata and Data Science DevRoom:
https://hpc-bigdata-fosdem17.github.io/
It was
What Till said is true for FlinkML, until all the moving parts are in place
there's not much point in annotating any as Public. The Spark project has
the @Experimental tag IIRC, that would fit our case better.
On Wed, Nov 23, 2016 at 4:09 PM, Till Rohrmann wrote:
> I think in general annotating
Hi Greg,
yes we could and as Aljoscha pointed out the proper solution would be to
serialize all objects (as done in the DataSet API) and not hold them as
objects on the heap.
This would be a major effort though and I am rather looking for a 'quick'
work around that does not have major side effects
Fabian Hueske created FLINK-5151:
Summary: Add discussion about object mutations to heap-based state
backend docs.
Key: FLINK-5151
URL: https://issues.apache.org/jira/browse/FLINK-5151
Project: Flink
On Wed, Nov 23, 2016 at 8:34 AM, Fabian Hueske wrote:
> The ReduceFunction must be used in the right way and it is easy to get
> wrong.
>
I'm likely highlighting my ignorance here, but if object reuse works
properly for ReduceFunction in the batch API, can we do the same in the
streaming API?
Chesnay Schepler created FLINK-5150:
---
Summary: WebUI metric-related resource leak
Key: FLINK-5150
URL: https://issues.apache.org/jira/browse/FLINK-5150
Project: Flink
Issue Type: Bug
I usually use apache commons CSV for that, as you can see here (inside the
*parseWithApacheCommonsCsv* part of the if):
https://github.com/okkam-it/flink-examples/blob/master/src/main/java/it/okkam/datalinks/batch/flink/datasourcemanager/importers/Csv2RowExample.java
I hope this could help!
Flavi
I think in general annotating library methods/classes is a good idea. The
question is just which APIs are going to be marked stable.
In the past we've seen that we might have marked some of Flink's APIs
stable too early. As a consequence we have to carry them along for quite
some time (at the very
Maximilian Michels created FLINK-5149:
-
Summary: ContinuousEventTimeTrigger doesn't fire at the end of the
window
Key: FLINK-5149
URL: https://issues.apache.org/jira/browse/FLINK-5149
Project: Fli
Chesnay Schepler created FLINK-5148:
---
Summary: LocalFileSystem#delete can fail with NullpointerException
Key: FLINK-5148
URL: https://issues.apache.org/jira/browse/FLINK-5148
Project: Flink
Hi,
I'm working on https://issues.apache.org/jira/browse/FLINK-2186
As I understand, Flink cannot read wide-column files in tuple, but pojo
So far we must create that pojo manually, it's convenient when count of columns
not so many
When it's over thousand - hardly seems possible
To solve this i
Yes, you're right. This is not a principled solution but rather a
work-around for a specific use case.
The ReduceFunction must be used in the right way and it is easy to get
wrong. (OTOH, there is currently no way to get object reusage right. So
think the change would not worsen the current state.
Chesnay Schepler created FLINK-5147:
---
Summary: StreamingOperatorsITCase.testGroupedFoldOperation failed
on Travis
Key: FLINK-5147
URL: https://issues.apache.org/jira/browse/FLINK-5147
Project: Flink
I would be for also annotating library methods/classes. Maybe Robert has a
stronger opinion on this because he introduced these annotations.
On Tue, 22 Nov 2016 at 18:56 Greg Hogan wrote:
> Hi all,
>
> Should stable APIs in Flink's CEP, ML, and Gelly libraries be annotated
> @Public or restricte
You can go ahead and do the change. I just think that this is quite
fragile. For example, this depends on the reduce function returning the
right object for reuse. If we hand in the copied object as the first input
and the ReduceFunction reuses the second input then we again have a
reference to the
Stefan Richter created FLINK-5146:
-
Summary: Improved resource cleanup in RocksDB keyed state backend
Key: FLINK-5146
URL: https://issues.apache.org/jira/browse/FLINK-5146
Project: Flink
Issu
Chesnay Schepler created FLINK-5145:
---
Summary: WebInterface to aggressive in pulling metrics
Key: FLINK-5145
URL: https://issues.apache.org/jira/browse/FLINK-5145
Project: Flink
Issue Type:
Hi Aljoscha,
sure, there many issues with holding the state as objects on the heap.
However, I think we don't have to solve all problems related to that in
order to add a small fix that solves one specific issue.
I would not explicitly expose the fix to users but it would be nice if we
could imple
Timo Walther created FLINK-5144:
---
Summary: Error while applying rule AggregateJoinTransposeRule
Key: FLINK-5144
URL: https://issues.apache.org/jira/browse/FLINK-5144
Project: Flink
Issue Type:
hiļ¼Fabian Hueske, Sorry for mistake for the whole PR #2792
> On Nov 23, 2016, at 17:10, Fabian Hueske wrote:
>
> Hi,
>
> Why do you think that this means "much code changes"?
> I think it would actually be a pretty lightweight change in
> HeapReducingState.
>
> The proposal is to copy the *fir
+1
This should always be the norm, especially for user-facing code.
While we are at it, perhaps when someone deprecates functionality the new
alternative should also be replaced right away.
E.g. Checkpointed is deprecated but all state management tests are actually
using this alternative.
chee
+1 and we should apply the same to all deprecated interfaces/abstract classes.
> On Nov 23, 2016, at 11:13 AM, Aljoscha Krettek wrote:
>
> +1 That sounds excellent.
>
> On Wed, 23 Nov 2016 at 11:04 Till Rohrmann wrote:
>
>> +1 for your proposal.
>>
>> Cheers,
>> Till
>>
>> On Wed, Nov 23, 2
+1 That sounds excellent.
On Wed, 23 Nov 2016 at 11:04 Till Rohrmann wrote:
> +1 for your proposal.
>
> Cheers,
> Till
>
> On Wed, Nov 23, 2016 at 9:33 AM, Fabian Hueske wrote:
>
> > I agree on this one.
> > Whenever we deprecate a method or a feature we should add a comment that
> > explains t
I think it's not a good idea to introduce this special case fix for one
specific use case because this can have implications for other parts of the
code. We should push for keeping the data in serialised form. There's also
other problems, for example, a ListState allows modifying the returned
Itera
+1 for your proposal.
Cheers,
Till
On Wed, Nov 23, 2016 at 9:33 AM, Fabian Hueske wrote:
> I agree on this one.
> Whenever we deprecate a method or a feature we should add a comment that
> explains the new API or why the feature was removed without replacement.
>
> Enforcing this information th
Timo Walther created FLINK-5143:
---
Summary: Add EXISTS to list of supported operators
Key: FLINK-5143
URL: https://issues.apache.org/jira/browse/FLINK-5143
Project: Flink
Issue Type: Bug
Frank Lauterwald created FLINK-5142:
---
Summary: Resource leak in CheckpointCoordinator
Key: FLINK-5142
URL: https://issues.apache.org/jira/browse/FLINK-5142
Project: Flink
Issue Type: Bug
Hi,
Why do you think that this means "much code changes"?
I think it would actually be a pretty lightweight change in
HeapReducingState.
The proposal is to copy the *first* value that goes into a ReducingState.
The copy would be done by a TypeSerializer and hence be a deep copy.
This will allow t
Biao Liu created FLINK-5141:
---
Summary: Implement MiniClusterStreamEnvironment to run new mini
cluster in flip-6 branch
Key: FLINK-5141
URL: https://issues.apache.org/jira/browse/FLINK-5141
Project: Flink
I agree on this one.
Whenever we deprecate a method or a feature we should add a comment that
explains the new API or why the feature was removed without replacement.
Enforcing this information through checkstyle makes sense as well, IMO.
Cheers, Fabian
2016-11-23 4:42 GMT+01:00 sjk :
> Hi, all
30 matches
Mail list logo