[Nepomuk] Fwd: GSoC 2012 - Testing and Benchmarking in Nepomuk

Sun Apr 8 07:27:26 UTC 2012

---------- Forwarded message ----------
From: Raveendra Bhat <ravee.bhat69 at gmail.com>
Date: Sun, Apr 8, 2012 at 12:57 PM
Subject: Re: [Nepomuk] GSoC 2012 - Testing and Benchmarking in Nepomuk
To: Vishesh Handa <me at vhanda.in>

Hi,

Coding period of GSoC starts from 21st may. But I have semister end exams
starting from 14th may and finishes on 26th may. I have mentioned this in
my proposal for Testing and Benchmarking in Nepomuk. So I wont be available
for thee first week of GSoC coding period.

On Wed, Apr 4, 2012 at 2:23 PM, Raveendra Bhat <ravee.bhat69 at gmail.com>wrote:

> Hi Vishesh,
>
> About my Qt and KDE programming experience:
>
> I am familiar with Qt desktop and mobile application development.
>
> I was working on developing a desktop application which aimed at network
> traffic monitoring and management. But the project was not feasible and it
> already exists.
>
> I have basic knowledge of using nepomuk in desktop applications and I am
> good with its metadata management, how it allows interconnections of data
> between  different desktop applications  etc.
>
> Now I am trying to understand DMS APIs and DBus architecture. And how DBus
> interface is used in Data Management Services.
>
>
>
>
> On Wed, Apr 4, 2012 at 1:13 AM, Vishesh Handa <me at vhanda.in> wrote:
>
>>
>> On Sun, Apr 1, 2012 at 12:30 PM, Raveendra Bhat <ravee.bhat69 at gmail.com>wrote:
>>
>>> Hi,
>>>
>>> Sorry for the late reply.I had my academic tests and  I was waiting for
>>> your reply for my last mail. In the meanwhile I have done following works:
>>>
>>> 1. First of all I cleared my misconception about benchmarks, that it is
>>> done only on two separate machines. And learnt that there are other
>>> circumstances where benchmarks are also used to measure performances of a
>>> function an operation etc...
>>>
>>> 2.Also I came across Sebastian Trueg's blog about why a central DBus
>>> architecture is required for nepomuk data management.
>>>
>>> 3. Studied about QTestLib framework and API. How a basic test is created
>>> and how QBENCHMARK macro is added to a test function that we want to
>>> benchmark.
>>>
>>> 4. Also tried few of the examples given in QTestLib tutorial and
>>> succeeded to write test functions for my own small sample class.
>>>
>>> Sorry, I couldn't do more work because of my college tests. I hope you
>>> would give me a feedback on the work i have done and what more work to be
>>> done. I believe it would help me come up with a proof of concept and
>>> writing a proposal.
>>>
>>>
>>> On Fri, Mar 30, 2012 at 12:50 AM, Raveendra Bhat <ravee.bhat69 at gmail.com
>>> > wrote:
>>>
>>>> Hi Vishesh,
>>>>
>>>> Firstly thanks for your replyI went through your blog post and had a
>>>> look at your code. I could able to understand your filewatcher test and
>>>> identificationtest. I have some doubts with respect to what needs to be
>>>> done in the GSoC period.
>>>>
>>>> 1. Project statement clearly says, I need to write testcases for
>>>> nepomuk services. Does it involve ontology,storage,query and strigi
>>>> services? What exactly do you mean by porting Nepomuk::Resources to the
>>>> testing framework? Writing testcases for all Nepomuk::Resource
>>>> properties/methods?
>>>>
>>>
>> Yup. In fact some of the tests already exist, but since Nepomuk::Resource
>> has been ported to the new architecture, it now relies on dbus and requires
>> a running nepomuk storage.
>>
>> Nepomuk::Storage does a lot of caching, and tries to keep the cache up to
>> date. That all needs tests.
>>
>>  2. According to my knowledge about bench marks, I believe that it
>>>> should be done on 2 seperate systems under test. Can you please give me a
>>>> clear picture of *benchmarking* the caching time,property fetch time?
>>>> 3.Bench marking for file indexer means some tool like system monitor
>>>> which displays memory and CPU usage of the indexer?
>>>>
>>>
>> yes. In fact I'd like something to monitor the memory and cpu usage of
>> virtuoso as well.
>>
>> Benchmarking strigi - Needs to be benchmarked (with memory and cpu usage)
>> for different file types
>> Benchmarking virtuoso - We need to be able to see how fast certain
>> queries run and how long much memory does virtuoso consume. This will help
>> a lot in the process of optimizing the sparql query libraries
>> (kdelibs/nepomuk/query)
>>
>> So, overall here is what I'm looking for -
>>
>> * Benchmarks for strigi and virtuoso
>> * Unit tests for Nepomuk::Resource
>> * Integrated testing for the file-watcher and indexer ( This will involve
>> creating a mock kde session and touching certain files to see if they are
>> re-indexed )
>>
>> Eventually I should be able to write a test for say 'storeResources' and
>> see how much memory and cpu virtuoso are consuming. Or compare between
>> different nepomuk versions to see how fast it is to push in large blobs of
>> data.
>>
>> Plus, we're going to be completely re-structuring the Nepomuk::Resource
>> internals, so your tests will go a long way in making sure that we do not
>> break anything.
>>
>> Do you have any prior experience with Qt or KDE?
>>
>>
>>>>
>>>> Waiting for your reply. Mean while I'll be looking at your code and
>>>> will come up with a proof of concept.
>>>>
>>>>
>>>> On Wed, Mar 28, 2012 at 8:26 PM, Vishesh Handa <me at vhanda.in> wrote:
>>>>
>>>>>
>>>>>
>>>>> On Wed, Mar 28, 2012 at 6:10 PM, Raveendra Bhat <
>>>>> ravee.bhat69 at gmail.com> wrote:
>>>>>
>>>>>> Hi,
>>>>>>
>>>>>>   I am Raveendra from India. I am interested in writing a Test
>>>>>> framework for nepomuk.  I have a basic knowledge how nepomuk works.I am
>>>>>> familiar with Qt C++ development.But i am not familiar with testing
>>>>>> libraries in Qt/C++.
>>>>>>
>>>>>> Please can you tell me more in details about this project? I want to
>>>>>> be a Kontributor.
>>>>>>
>>>>>
>>>>> Hey Raveendra
>>>>>
>>>>> I'm basically expecting someone to continue with my test framework
>>>>> [1]. That would involved porting the Nepomuk::Resource tests to the test
>>>>> framework, cause they now require a dbus session.
>>>>>
>>>>> Additionally, I would want benchmarks on Nepomuk::Resource. How long
>>>>> does it take to fill up the cache? Fetching properties and so on. You'll
>>>>> even need to write more tests for it.
>>>>>
>>>>> Now with the introduction with Nepomuk 2.0 and the data management
>>>>> API, I would want benchmarks on the new functions as well. ( They already
>>>>> have a lot of unit tests, so you do not need to write those )
>>>>>
>>>>> I guess, I'd also want some kind of benchmarks for the file indexer.
>>>>>
>>>>> That's just the start. Look at every existing nepomuk service. If they
>>>>> do not have tests, they need them.
>>>>>
>>>>> [1] http://vhanda.in/blog/2012/03/nepomuk-test-framework/
>>>>>
>>>>>>
>>>>>> --
>>>>>> regards,
>>>>>>
>>>>>> B R Raveendra
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Nepomuk mailing list
>>>>>> Nepomuk at kde.org
>>>>>> https://mail.kde.org/mailman/listinfo/nepomuk
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Vishesh Handa
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> regards,
>>>>
>>>> B R Raveendra
>>>>
>>>>
>>>
>>>
>>> --
>>> regards,
>>>
>>> B R Raveendra
>>>
>>>
>>
>>
>> --
>> Vishesh Handa
>>
>>
>
>
> --
> regards,
>
> B R Raveendra
>
>

-- 
regards,

B R Raveendra

-- 
regards,

B R Raveendra
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/nepomuk/attachments/20120408/2acdf70d/attachment.html>