Mailing-List: contact common-dev-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: common-dev@hadoop.apache.org
Received-SPF: neutral (athena.apache.org: local policy)
DomainKey-Signature: a=rsa-sha1; s=serpent; d=yahoo-inc.com; c=nofws; q=dns;
	h=message-id:date:from:user-agent:mime-version:to:subject:
	references:in-reply-to:content-type:content-transfer-encoding;
	b=MfFVbEOhl2SPieP9MIKvlJV4oQxGH1vh2IB80Yyx3bOvW9ICz5W6vcLcefMTlMcd
Message-ID: <4BEDB06C.6060400@yahoo-inc.com>
Date: Fri, 14 May 2010 13:19:56 -0700
From: Konstantin Shvachko <shv@yahoo-inc.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US;
 rv:1.9.1.9) Gecko/20100317 Thunderbird/3.0.4
MIME-Version: 1.0
To: common-dev@hadoop.apache.org
Subject: Re: [DISCUSSION]: Integrating SureLogic into Hadoop
References: <20100506015053.GP28975@goodenter-lm.boudnik.org>
 <k2l45f85f71005071635gd43365bbr8efdfe1d797078f7@mail.gmail.com>
 <20100510233715.GE68237@goodenter-lm.boudnik.org>
 <20100514195947.GM94793@goodenter-lm.boudnik.org>
In-Reply-To: <20100514195947.GM94793@goodenter-lm.boudnik.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

This is good news!
I found SureLogic stack useful for finding bugs.
It was especially helpful in detecting synchronization issues.
Good that the licensing issues are cleared out.
Thanks, Cos.
--Konstantin


On 5/14/2010 12:59 PM, Konstantin Boudnik wrote:
> Here's an update from SureLogic on the licensing of the software to the
> broader contributors community.
>
> 1) For now we should be able to use 'committers' license (the one we already have)
>     to share with contributors on per case basis (a contributor needs to ask for it and after
>     a consideration by a designated committer it can be send our to the
>     requester)
> 2) in the future i.e. past 10/31/2010 when the current license will expire and
>     be updated, we'll provide a POC so that our contributors can ask for a
>     license to be given to them (similar to #1 above).
>
> Right now, SureLogic works on license auto-update feature where new license
> will be automatically retrieved by the tools upon the expiration of an old
> one. Thus #2) will be moving to that "self-renewing" license mechanism to
> avoid having to pass out new licenses each year.
>
> So, I think this all seems very reasonable and we can effectively close the
> 'license' issue.
>
> On the other hand, I got the word that CLI interfaces for their tools are in
> the testing phase right now, so we should be able to try them out shortly.
>
> Do we have any other concerns at this point?
>
> Cos
>
> On Mon, May 10, 2010 at 04:37PM, Konstantin Boudnik wrote:
>> This is very valid concern Todd. I am talking with SureLogic at the moment to
>> find out if they will be willing to provide the license to all Hadoop
>> contributors.
>>
>> On the other hand not including such type of validation into standard patch
>> validation process poses a danger of code degradation as time passes. E.g. we
>> can add SureLogic validation into secondary build but then Hadoop committers
>> will have to make sure that no new SureLogic's warnings were added in the
>> last week or so and fix them immediately or at least open a JIRA to track such
>> issues. Certainly there some other ways.
>>
>> I believe we might need to make an effort here to help SureLogic to limit the
>> distribution of the license by Hadoop contributors crowd. I am not sure if this is
>> Ok to ask contributors to fill some kind of legal form where they agree not to
>> use the copy of the license for any projects not hosted by ASF or something?
>> Similar to what we fill for our the contributions to ASF? But first let's hear
>> SureLogic take on the license ;)
>>
>> Cos
>>
>> On Fri, May 07, 2010 at 04:35PM, Todd Lipcon wrote:
>>> Hi Cos,
>>>
>>> This looks great, and I'm excited to have more ways of finding these tricky
>>> bugs. Are there any examples of bugs found already by these techniques?
>>>
>>> The one concern I have about the proposal is with this:
>>>> SureLogic analysis is going to be included to the test-patch process. This
>>> said new patches are required not to raise SureLogic warnings level (similar
>>> to the requirements about FindBugs or javac).
>>>
>>> This is slightly worrisome since the SureLogic license is only available to
>>> committers. For non-committers, I think this may prove to be difficult since
>>> they won't have any local means of checking for warnings and verifying that
>>> they've fixed them.
>>>
>>> Thanks
>>> -Todd
>>>
>>>
>>> 2010/5/5 Konstantin Boudnik<cos@yahoo-inc.com>
>>>
>>>> Hello.
>>>>
>>>> As some of you might know we have the license for great concurrency
>>>> analysis
>>>> software from SureLogic.
>>>>
>>>> SureLogic engineers gave some run for HDFS, MR, and Zookeeper code at the
>>>> end
>>>> of last year. The tools seem very promising and supposedly bring us the
>>>> value (linked from the page below).
>>>>
>>>> Please read this Wiki page to get more information about the tool and to
>>>> get some understanding how it works
>>>>   http://wiki.apache.org/hadoop/HowToUseConcurrencyAnalysisTools
>>>>
>>>> Here's the additional information on how to install/upgrade to the latest
>>>> release: http://surelogic.com/static/eclipse/install.html
>>>>
>>>> There's also a couple of JIRAs w/ patches (HDFS-801, MAPREDUCE-1259) to
>>>> integrate annotations into our current source code base. The annotations
>>>> are
>>>> represented as a jar file with some interface classes. They are
>>>> redistributed
>>>> under Apache license. The retention policy of annotations are 'compile
>>>> time',
>>>> i.e. this jar isn't required for the runtime of Hadoop.
>>>>
>>>> It'd be great to hear community thoughts on this so we can make some
>>>> decision
>>>> about this toolset.
>>>>
>>>> Please comment. Thanks
>>>>   Cos
>>>>
>>>
>>>
>>>
>>> --
>>> Todd Lipcon
>>> Software Engineer, Cloudera