Andra Lungu created FLINK-2310:
--
Summary: Add an Adamic-Adar Similarity example
Key: FLINK-2310
URL: https://issues.apache.org/jira/browse/FLINK-2310
Project: Flink
Issue Type: Task
Co
Hi all
I'm trying to work out a general convergence framework for Machine Learning
Algorithms which utilize iterations for optimization. For now, I can think
of three kinds of convergence functions which might be useful.
1. converge(data, modelBeforeIteration, modelAfterIteration)
2. converge(data,
Stephan Ewen created FLINK-2309:
---
Summary: Use static code analysis to replace selector functions
with field accesses.
Key: FLINK-2309
URL: https://issues.apache.org/jira/browse/FLINK-2309
Project: Flin
Hi everybody,
does anybody know whether there is an implementation of the
hadoopinputformats for matrices in numpy or .mat format? This would be
really needed since a lot of machine learning data is stored in these
formats.
Thanks for your help,
Felix
+1 on offering both.
This way we cover both simplicity and expressivity when needed. Annotations
give a very clean and simple way for marking state imho.
Perhaps most casual users will find it much better to just tag the fields that
they want to persist.
Paris
> On 01 Jul 2015, at 15:55, Step
Stephan Ewen created FLINK-2308:
---
Summary: Improve error messages on broken user-defined
serialization
Key: FLINK-2308
URL: https://issues.apache.org/jira/browse/FLINK-2308
Project: Flink
Issu
Actually, this is the first rework of the state interface. There is only
one released version, yet.
What we are doing here is trying to make sure that this first rework will
most likely also be the last for the foreseeable future.
>From the use cases I can think of, we need at least two different
Robert Metzger created FLINK-2307:
-
Summary: Drop Java 6 support
Key: FLINK-2307
URL: https://issues.apache.org/jira/browse/FLINK-2307
Project: Flink
Issue Type: Improvement
Compone
I created a JIRA for this: https://issues.apache.org/jira/browse/FLINK-2307
On Wed, Jun 24, 2015 at 3:01 PM, Till Rohrmann wrote:
> +1
>
> On Tue, Jun 23, 2015 at 3:16 PM, Robert Metzger
> wrote:
>
> > +1
> >
> >
> >
> > On Tue, Jun 23, 2015 at 11:31 AM, Fabian Hueske
> wrote:
> >
> > > +1
> >
I agree, if we want to change the interface, now is the best time.
So you are suggesting to change the methods in the Checkpointed interface
from
T snapshotState(long checkpointId, long checkpointTimestamp) throws Exception;
void restoreState(T state);
to
void onSnapshot(id, ts)
void onRestore
I understand your concerns Robert but I don't fully agree.
The Checkpointed interface works indeed but there are so many use cases
that it is not suitable for in the long run, and also the whole interface
is slightly awkward in my opinion when returning simple fields which are
already serializable
Matthias J. Sax created FLINK-2306:
--
Summary: Add support for named streams in Storm compatibility layer
Key: FLINK-2306
URL: https://issues.apache.org/jira/browse/FLINK-2306
Project: Flink
Matthias J. Sax created FLINK-2305:
--
Summary: Add documenation about Storm compatibility layer
Key: FLINK-2305
URL: https://issues.apache.org/jira/browse/FLINK-2305
Project: Flink
Issue Type
Matthias J. Sax created FLINK-2304:
--
Summary: Add named attribute access to Storm compatibility layer
Key: FLINK-2304
URL: https://issues.apache.org/jira/browse/FLINK-2304
Project: Flink
Iss
Whats causing me the biggest headache here is that I don't see an end on
all these "state interface" reworks.
I think this is now the third big change to the interface.
It is a horrible user experience to rework your old code with each new
Flink release.
I understand that there are always ways to
On 01 Jul 2015, at 10:57, Gyula Fóra wrote:
> Hey,
>
> Thanks for the feedback guys:
>
> @Max: You are right, this is not top priority to changes, I was just
> mocking up some alternatives to try to make the state usage even simpler so
> that the user can keep his current implementations and j
Gábor Hermann created FLINK-2303:
Summary: Add count method to GroupedDataStream
Key: FLINK-2303
URL: https://issues.apache.org/jira/browse/FLINK-2303
Project: Flink
Issue Type: Improvement
Hey,
Thanks for the feedback guys:
@Max: You are right, this is not top priority to changes, I was just
mocking up some alternatives to try to make the state usage even simpler so
that the user can keep his current implementations and just add 1-2
annotations.
@Stephan, Robert: You are right tha
On 01 Jul 2015, at 10:34, Stephan Ewen wrote:
> +1, like that approach
+1
I like that this is not breaking for non-Scala users :-)
+1, like that approach
On Wed, Jul 1, 2015 at 10:28 AM, Robert Metzger wrote:
> (adding dev@ to the conversation)
>
> Chiwan looked into the issue. It seems that we can not add the Scala
> version only to flink-scala, flink-streaming-scala,
> Since flink-runtime also needs scala all modules
(adding dev@ to the conversation)
Chiwan looked into the issue. It seems that we can not add the Scala
version only to flink-scala, flink-streaming-scala,
Since flink-runtime also needs scala all modules are affected by this.
I would vote for naming the Scala 2.10 version of flink modules wi
+1 for adding the annotation, but not removing the interface
Robert is right, the nice thing about the current interface is that you can
use it to commit the state yourself to a database and simply return a key
to where the state is stored. That is quite nice.
On Wed, Jul 1, 2015 at 10:14 AM, Rob
I would certainly not replace the current Checkpointed interface by this
implementation.
The interface allows you to perform custom actions when creating a snapshot
or restoring state.
We could add the annotation variant for simple cases like in your example.
On Wed, Jul 1, 2015 at 10:10 AM, Max
Hi Gyula,
Looks like a neat feature you thought of; I like it. One problem I see is
that with all the big changes and discussions in streaming, I wonder if we
should get other things right first. Another problem could be the lack of
clarity of this implementation. State annotations can now be very
24 matches
Mail list logo
