You can use Jira filters to narrow down the scope of issues you want to
possible address, for instance, I use this filter to look into open issues,
that are unassigned :
https://issues.apache.org/jira/issues/?filter=12333428
For a specific release, you can also filter the release, and I Reynold
I just added snapshot builds for 1.5. They will take a few hours to
build, but once we get them working should publish every few hours.
https://amplab.cs.berkeley.edu/jenkins/view/Spark-Packaging
- Patrick
On Mon, Sep 21, 2015 at 10:36 PM, Bin Wang wrote:
> However I find
ok, here's the updated downtime schedule for this week:
wednesday, sept 23rd:
firewall maintenance cancelled, as jon took care of the update
saturday morning while we were bringing jenkins back up after the colo
fire
thursday, sept 24th:
jenkins maintenance is still scheduled, but abbreviated
Thanks for the links (first one is broken or private).
I think the main mistake I was making was looking at fix version instead of
target version (JIRA homepage with listings of versions links to fix
versions).
For anyone else interested in MLlib things, I am looking at this to see
what goals
I am puzzled by the behavior of column identifiers in Spark SQL. I don't
find any guidance in the "Spark SQL and DataFrame Guide" at
http://spark.apache.org/docs/latest/sql-programming-guide.html. I am seeing
odd behavior related to case-sensitivity and to delimited (quoted)
identifiers.
Are you using a SQLContext or a HiveContext? The programming guide
suggests the latter, as the former is really only there because some
applications may have conflicts with Hive dependencies. SQLContext is case
sensitive by default where as the HiveContext is not. The parser in
HiveContext is
HiveQL uses `backticks` for quoted identifiers.
On Tue, Sep 22, 2015 at 1:06 PM, Richard Hillegas
wrote:
> Thanks for that tip, Michael. I think that my sqlContext was a raw
> SQLContext originally. I have rebuilt Spark like so...
>
> sbt/sbt -Phive assembly/assembly
>
>
As Rui says it would be good to understand the use case we want to
support (supporting CRAN installs could be one for example). I don't
think it should be very hard to do as the RBackend itself doesn't use
the R source files. The RRDD does use it and the value comes from
Which Spark release are you building ?
For master branch, I get the following:
lib_managed/jars/datanucleus-api-jdo-3.2.6.jar
lib_managed/jars/datanucleus-core-3.2.10.jar
lib_managed/jars/datanucleus-rdbms-3.2.9.jar
FYI
On Tue, Sep 22, 2015 at 1:28 PM, Richard Hillegas
I see that lib_managed/jars holds these old Derby versions:
lib_managed/jars/derby-10.10.1.1.jar
lib_managed/jars/derby-10.10.2.0.jar
The Derby 10.10 release family supports some ancient JVMs: Java SE 5 and
Java ME CDC/Foundation Profile 1.1. It's hard to imagine anyone running
Spark on
Thanks for that tip, Michael. I think that my sqlContext was a raw
SQLContext originally. I have rebuilt Spark like so...
sbt/sbt -Phive assembly/assembly
Now I see that my sqlContext is a HiveContext. That fixes one of the
queries. Now unnormalized column names work:
// ...unnormalized
Hi Devs,
Hopefully one of you know more on this?
Thanks
Andy
-- Forwarded message --
From: Andy Huang
Date: Wed, Sep 23, 2015 at 12:39 PM
Subject: Parallel collection in driver programs
To: u...@spark.apache.org
Hi All,
Would like know if anyone
I see.
I use maven to build so I observe different contents under lib_managed
directory.
Here is snippet of dependency tree:
[INFO] | +- org.spark-project.hive:hive-metastore:jar:1.2.1.spark:compile
[INFO] | | +- com.jolbox:bonecp:jar:0.8.0.RELEASE:compile
[INFO] | | +-
Thanks, Ted. I'll follow up with the Hive folks.
Cheers,
-Rick
Ted Yu wrote on 09/22/2015 03:41:12 PM:
> From: Ted Yu
> To: Richard Hillegas/San Francisco/IBM@IBMUS
> Cc: Dev
> Date: 09/22/2015 03:41 PM
> Subject: Re: Derby
Thanks for that additional tip, Michael. Backticks fix the problem query in
which an identifier was transformed into a string literal. So this works
now...
// now correctly resolves the unnormalized column id
sqlContext.sql("""select `b` from test_data""").show
Any suggestion about how to
Thanks, Ted. I'm working on my master branch. The lib_managed/jars
directory has a lot of jarballs, including hadoop and hive. Maybe these
were faulted in when I built with the following command?
sbt/sbt -Phive assembly/assembly
The Derby jars seem to be used in order to manage the
I cloned Hive 1.2 code base and saw:
10.10.2.0
So the version used by Spark is quite close to what Hive uses.
On Tue, Sep 22, 2015 at 3:29 PM, Ted Yu wrote:
> I see.
> I use maven to build so I observe different contents under lib_managed
> directory.
>
> Here is
Hi,
Recently,I am reading source code(1.5 version) about sparksql .
In DataFrame.scala, there is a funtion named filter in the 737 row
*def filter(condition: Column): DataFrame = Filter(condition.expr,
logicalPlan)*
The fucntion return a Filter object,but it require a DataFrame
Thanks. I've solved it. I modified pom.xml and add my own repo into it,
then use "mvn deploy".
Fengdong Yu 于2015年9月22日周二 下午2:08写道:
> basically, you can build snapshot by yourself.
>
> just clone the source code, and then 'mvn package/deploy/install…..’
>
>
> Azuryy Yu
>
Hi all,
wondering if any could make the new 1.5.0 stallSkinnyQR to work.
Follows my output, which is a big loop of the same errors until the shell dies.
I am curious since im failing to load any implementations from BLAS, LAPACK,
etc.
scala> mat.tallSkinnyQR(false)
15/09/22 10:18:11 WARN
Where is the best place to look at open issues that haven't been
assigned/started for the next release? I am interested in working on
something, but I don't know what issues are higher priority for the next
release.
On a similar note, is there somewhere which outlines the overall goals for
the
21 matches
Mail list logo