airavata-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Suresh Marru <>
Subject Re: Airavata GSoC 2013 Master Project
Date Sun, 14 Apr 2013 19:05:54 GMT
Hi Andun and others (i am sure every one has identical concerns):

I am not intending everything within the wiki has to be accomplished by one. They are easily
4 projects and may be more. But I want you all to understand everything on the wiki because
they are highly inter-related. For instance you may feel a certain way of composing workflow
may be better, but you need to understand how the composed workflow will be used. Understanding
experiment builder concept will help what should be within the workflow graph and so on. 

You are right, implementing a component will require you to do a proper related study and
so forth. Once all of you have the big picture in mind, you can dive into a particular implementation
being cognizant of how it will be integrated. I am insisting all of this is happening on the
mailing lists with incremental contributions. 

This process certainly is complicated than a typical gsoc project, but as I said before, this
is to mitigate projects executed in isolation which do not shape back to main code. And secondly
these are issues most of the airavata dev's are working with, so you will get good mentor

I put out the picture but did not describe individual steps, this is intentional. I did not
want to prescriptive on how it has to be. I just wanted to be suggestive and open up to your

Also note that, if you all convince well, everything can be replaced -- including all the
data models and service interfaces. But you will need to have good justifications. I for myself
will not be convinced if you say it is a cool thing to do or it is a popular or buzzy thing.
I will certainly be more than convinced if you reason well how it will help address current
issues and will open up for future extensions. 


On Apr 14, 2013, at 2:11 PM, Andun Sameera <> wrote:

> Hi Suresh,
> I have read the GSOC 2013 wiki article. That describes a big process,
> which have to be followed very carefully. So I want to clear myself.
> As you mentioned in the wiki, perfect understanding Data Models and
> Service Interfaces is mandatory for a GSOC project. There are four
> major components in the Web based UI(Which are shown in the diagram).
> According to my knowledge, the generic task of all these component is
> to manipulate data models and use service interfaces to achieve the
> intended goal of  the particular component.
> For an example, The component Application Registering UI should get
> some inputs from user about inputs out puts etc. about a application.
> Then it have to map those to ServiceMapSchema to create an Application
> and Register it in Airavata via Service Interfaces.
> Also as I understood Data Models and Service Interfaces are fixed
> things and we have to use these in our projects. But sometimes changes
> can happen. That is the point where four separate components
> implementation project have to communicate. Otherwise if we only have
> to use the exiting Data Models and Service Interfaces to implement
> these web based components, we can work separate.
> To divide these four major components to GSOC projects, we have to
> understand the related work load for each of the implementations.
> Sometimes these can be four project, more than four or less.
> Thanks!
> On Sun, Apr 14, 2013 at 10:56 PM, Suresh Marru <> wrote:
>> Hi All,
>> Also please do not get too carries away with workflow description. Airavata's key
goal is to provide abstraction over distributed computing applications.  The decisions on
what application description should look like should be based on if it can cover all the intended
use cases. The Open Grid Forum's JSDL's specification will be a good reference for skimming
through -
>> Suresh
>> On Apr 14, 2013, at 9:35 AM, Suresh Marru <> wrote:
>>> Hi All,
>>> As briefly discussed in [1], lets brainstorm enough on this thread before we
can derive concrete GSoC projects and appropriate JIRA's. We have an opportunity to have multiple
projects, but we would like to make sure they all align well. I drafted a overview of what
can be potentially done - [2]. Please do not get overwhelmed by the sheer of changes, at the
same time, it will be time consuming and also will require co-ordination with the community
and other students. So please read [1] and make sure it matches your interest.
>>> Please start asking questions based on what you understand in [2] and lets go
from there. I will keep expanding this wiki as we get the questions. We are roughly 2 weeks
away from the proposal deadlines, so please hurry and understand the problem to tackle well.
Then lets collectively identify the project which is "doable" within next 3 months and how
we can break it down. If this is sounding too onerous for a GSoC project, I frankly suggest
to look at other projects which might fit your time availability and get the project. Here
we would like to use the mentor energy to accomplish tasks which can directly get into the
main code base, write papers, see the projects in production and so on. The rewards will be
certainly better than an average project, but the challenge and effort will be demanding.
>>> I will be relatively very responsive from now through proposal deadlines. I will
hope others on the Airavata PMC will be as well.
>>> Thanks,
>>> Suresh
>>> [1] -
>>> [2] -
> -- 
> Regards
> Andun S.L. Gunawardana
> Undergraduate
> Department of Computer Science And Engineering
> University of Moratuwa
> Sri Lanka
> Blog -
> LinkedIn -
> Twitter -

View raw message