You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@airavata.apache.org by Marlon Pierce <ma...@iu.edu> on 2014/02/03 15:14:18 UTC
Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Hi all--
We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
URL will be forthcoming.
Marlon
Re: Summary Re: Hangout: Orchestrator, Registry, GFAC, and API recap
and plans
Posted by Marlon Pierce <ma...@iu.edu>.
Please see the Thrift definitions in modules/thrift-interfaces.
Marlon
On 2/3/14 11:14 PM, Nipun Udara wrote:
> Hi Suresh,
>
> Can you let me know what are the new data model Objects defined in API that
> mentioned above. Thank you
>
> Regards
>
>
> On Tue, Feb 4, 2014 at 1:32 AM, Suresh Marru <sm...@apache.org> wrote:
>
>> Thanks Marlon for a comprehensive yet concise summary.
>>
>> For the API Server, how about Airavata API Facade - drawing from
>> http://en.wikipedia.org/wiki/Facade_pattern
>>
>> This does not make a good acronym either. So better suggestions will be
>> great.
>>
>> Suresh
>>
>> On Feb 3, 2014, at 2:54 PM, Marlon Pierce <ma...@iu.edu> wrote:
>>
>>> The following will be associated with specific Jira tasks. Here is the
>>> summary:
>>>
>>> * Suresh is done with the Thrift API for now, ready from critical
>> feedback.
>>> * Registry needs to implement the data model objects defined in Suresh's
>>> API and map these to database schema and OpenJPA.
>>>
>>> * Registry needs to define its Component Programming Interface (CPI).
>>> Ideally, this will be just "get" and "set" methods associated with
>>> Thrift-defined data model objects.
>>>
>>> * Orchestrator and GFAC components both also need CPIs. CPIs are Java
>>> interfaces with ".cpi." in the name space. This is refactoring current
>>> code for the most part.
>>>
>>> * While waiting on the Registry changes, we need to evaluate the Thrift
>>> data models from the point of view of the Orchestrator and GFAC.
>>>
>>> * We also need to evaluate the Thrift API from the point of view of
>>> external gateways.
>>>
>>> * We will need to implement and test the Airavata API Server (or AAS, to
>>> give it a name and abbreviation). This is the Thrift server that maps
>>> API methods to CPI calls. External gateways call the API. The AAS maps
>>> these calls to the internal CPI invocations (to the Registry or
>>> Orchestrator).
>>>
>>> We also will need to do steps for the Workflow Interpreter but this will
>>> be discussed more later.
>>>
>>>
>>> Marlon
>>>
>>>
>>> On 2/3/14 1:04 PM, Marlon Pierce wrote:
>>>> Hangout URL is
>>>> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
>>>>
>>>>
>>>> Marlon
>>>>
>>>> On 2/3/14 9:14 AM, Marlon Pierce wrote:
>>>>> Hi all--
>>>>>
>>>>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
>>>>> URL will be forthcoming.
>>>>>
>>>>>
>>>>> Marlon
>>>>>
>>
Re: Summary Re: Hangout: Orchestrator, Registry, GFAC, and API recap
and plans
Posted by Nipun Udara <y....@gmail.com>.
Hi Suresh,
Can you let me know what are the new data model Objects defined in API that
mentioned above. Thank you
Regards
On Tue, Feb 4, 2014 at 1:32 AM, Suresh Marru <sm...@apache.org> wrote:
> Thanks Marlon for a comprehensive yet concise summary.
>
> For the API Server, how about Airavata API Facade - drawing from
> http://en.wikipedia.org/wiki/Facade_pattern
>
> This does not make a good acronym either. So better suggestions will be
> great.
>
> Suresh
>
> On Feb 3, 2014, at 2:54 PM, Marlon Pierce <ma...@iu.edu> wrote:
>
> > The following will be associated with specific Jira tasks. Here is the
> > summary:
> >
> > * Suresh is done with the Thrift API for now, ready from critical
> feedback.
> >
> > * Registry needs to implement the data model objects defined in Suresh's
> > API and map these to database schema and OpenJPA.
> >
> > * Registry needs to define its Component Programming Interface (CPI).
> > Ideally, this will be just "get" and "set" methods associated with
> > Thrift-defined data model objects.
> >
> > * Orchestrator and GFAC components both also need CPIs. CPIs are Java
> > interfaces with ".cpi." in the name space. This is refactoring current
> > code for the most part.
> >
> > * While waiting on the Registry changes, we need to evaluate the Thrift
> > data models from the point of view of the Orchestrator and GFAC.
> >
> > * We also need to evaluate the Thrift API from the point of view of
> > external gateways.
> >
> > * We will need to implement and test the Airavata API Server (or AAS, to
> > give it a name and abbreviation). This is the Thrift server that maps
> > API methods to CPI calls. External gateways call the API. The AAS maps
> > these calls to the internal CPI invocations (to the Registry or
> > Orchestrator).
> >
> > We also will need to do steps for the Workflow Interpreter but this will
> > be discussed more later.
> >
> >
> > Marlon
> >
> >
> > On 2/3/14 1:04 PM, Marlon Pierce wrote:
> >> Hangout URL is
> >> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> >>
> >>
> >> Marlon
> >>
> >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> >>> Hi all--
> >>>
> >>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> >>> URL will be forthcoming.
> >>>
> >>>
> >>> Marlon
> >>>
> >
>
>
Re: Summary Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Suresh Marru <sm...@apache.org>.
Thanks Marlon for a comprehensive yet concise summary.
For the API Server, how about Airavata API Facade - drawing from http://en.wikipedia.org/wiki/Facade_pattern
This does not make a good acronym either. So better suggestions will be great.
Suresh
On Feb 3, 2014, at 2:54 PM, Marlon Pierce <ma...@iu.edu> wrote:
> The following will be associated with specific Jira tasks. Here is the
> summary:
>
> * Suresh is done with the Thrift API for now, ready from critical feedback.
>
> * Registry needs to implement the data model objects defined in Suresh's
> API and map these to database schema and OpenJPA.
>
> * Registry needs to define its Component Programming Interface (CPI).
> Ideally, this will be just "get" and "set" methods associated with
> Thrift-defined data model objects.
>
> * Orchestrator and GFAC components both also need CPIs. CPIs are Java
> interfaces with ".cpi." in the name space. This is refactoring current
> code for the most part.
>
> * While waiting on the Registry changes, we need to evaluate the Thrift
> data models from the point of view of the Orchestrator and GFAC.
>
> * We also need to evaluate the Thrift API from the point of view of
> external gateways.
>
> * We will need to implement and test the Airavata API Server (or AAS, to
> give it a name and abbreviation). This is the Thrift server that maps
> API methods to CPI calls. External gateways call the API. The AAS maps
> these calls to the internal CPI invocations (to the Registry or
> Orchestrator).
>
> We also will need to do steps for the Workflow Interpreter but this will
> be discussed more later.
>
>
> Marlon
>
>
> On 2/3/14 1:04 PM, Marlon Pierce wrote:
>> Hangout URL is
>> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
>>
>>
>> Marlon
>>
>> On 2/3/14 9:14 AM, Marlon Pierce wrote:
>>> Hi all--
>>>
>>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
>>> URL will be forthcoming.
>>>
>>>
>>> Marlon
>>>
>
Summary Re: Hangout: Orchestrator, Registry, GFAC, and API recap
and plans
Posted by Marlon Pierce <ma...@iu.edu>.
The following will be associated with specific Jira tasks. Here is the
summary:
* Suresh is done with the Thrift API for now, ready from critical feedback.
* Registry needs to implement the data model objects defined in Suresh's
API and map these to database schema and OpenJPA.
* Registry needs to define its Component Programming Interface (CPI).
Ideally, this will be just "get" and "set" methods associated with
Thrift-defined data model objects.
* Orchestrator and GFAC components both also need CPIs. CPIs are Java
interfaces with ".cpi." in the name space. This is refactoring current
code for the most part.
* While waiting on the Registry changes, we need to evaluate the Thrift
data models from the point of view of the Orchestrator and GFAC.
* We also need to evaluate the Thrift API from the point of view of
external gateways.
* We will need to implement and test the Airavata API Server (or AAS, to
give it a name and abbreviation). This is the Thrift server that maps
API methods to CPI calls. External gateways call the API. The AAS maps
these calls to the internal CPI invocations (to the Registry or
Orchestrator).
We also will need to do steps for the Workflow Interpreter but this will
be discussed more later.
Marlon
On 2/3/14 1:04 PM, Marlon Pierce wrote:
> Hangout URL is
> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
>
>
> Marlon
>
> On 2/3/14 9:14 AM, Marlon Pierce wrote:
>> Hi all--
>>
>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
>> URL will be forthcoming.
>>
>>
>> Marlon
>>
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Suresh Marru <sm...@apache.org>.
Thanks Supun, these indeed are slick examples. And yes I agree Airavata is not hadoop and we should not require to install thrift to build airavata. We need to have them in separate profile and lets hope we rarely have to use it (it all depends on how quickly we can stabilize thrift IDL’s).
Suresh
On Feb 4, 2014, at 11:49 AM, Supun Kamburugamuva <su...@gmail.com> wrote:
> Hi Suresh,
>
> Here are couple of examples I had about using a script to generate code for Thrift
>
> http://git-wip-us.apache.org/repos/asf/accumulo.git
> You can find the code generation script under core/src/main/scripts
>
> Another simple project is
> https://github.com/yahoo/storm-yarn
>
> Hadoop is using prtotocol buffers and they require protocol buffers to be installed in the system to compile hadoop. May be Hadoop can afford to do that :)
>
> Thanks,
> Supun..
>
>
> On Tue, Feb 4, 2014 at 11:36 AM, Suresh Marru <sm...@apache.org> wrote:
> Thanks Supun. Do you have any good references of public repos which you liked the thrift organization? I am trying to follow how cassandra is doing it. We also need to worry about license headers to the generated code to pass the release checks.
>
> Suresh
>
> On Feb 4, 2014, at 11:28 AM, Supun Kamburugamuva <su...@gmail.com> wrote:
>
> > I think the general practice in Thrift is to include a script that can do the thrift source generation if it needs to be done. The generated code is committed to the source repo.
> >
> > You can also use the maven plugin along with a maven profile to generate the code, if code generation is required.
> >
> > Thanks,
> > Supun..
> >
> >
> > On Tue, Feb 4, 2014 at 11:21 AM, Suresh Marru <sm...@apache.org> wrote:
> > Frankly we need some first hand experience and keep iterating on approaches on how all of this will play out across versions. Lets start with some approach and go from there.
> >
> > Thrift has a maven plugin in development but Chathuri confirmed even it will need thrift to be installed locally. It makes sense since the thrift generator is in C++ and is platform dependent.
> >
> > Suresh
> >
> > On Feb 4, 2014, at 11:13 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> >
> > > Hi Suresh,
> > >
> > > Does thrift has a maven plugin to generate classes ? We can have interfaces in one module and different implementations in a separate module. If there's a change (API level change, the implementation modules should be changed, otherwise they shouldn't be compiled). But if someone wants the old one they can have a dependency to the old module and should be able to compile the impl module).
> > >
> > > Suresh, WDYT ?
> > >
> > > Regards
> > > Lahiru
> > >
> > >
> > > On Tue, Feb 4, 2014 at 11:10 AM, Suresh Marru <sm...@apache.org> wrote:
> > > On Feb 4, 2014, at 9:59 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> > >
> > > > Hi Suresh,
> > > >
> > > > Current airavataAPI looks very clean and easy to understand. Thanks to your very detail comments I was able to understand the process.
> > > >
> > > > I think we need to make sure we give a simple easy to use experimentModel.ConfigurationData object to configure. What is our plan with implementing experimentModel.ConfigurationData ?
> > >
> > > Yes, we need every one’s critical eye in making these configuration data to cover all the use cases, yet be simple and as concise as possible.
> > >
> > > For the implementation, thrift has a generator for java beans (in addition to standard java) which essentially makes all members to private with setters to return void. We can have this generated beans as a separate airavata-thrift-datamodel.jar and all components will use this the same way we now use workflow execution context or workflow tracking library. Not sure if this is what you asked.
> > >
> > > > Regards
> > > > Lahiru
> > > >
> > > >
> > > > On Tue, Feb 4, 2014 at 9:35 AM, Marlon Pierce <ma...@iu.edu> wrote:
> > > > Where do these go? In modules/thrift-interfaces?
> > >
> > > I am now poking around at other thrift based projects to find a way to organize these. A challenge I see if how to over-write the generated classes without loosing any implementations. Will post a layout for discussion soon.
> > >
> > > Suresh
> > >
> > >
> > > >
> > > > Marlon
> > > >
> > > > On 2/4/14 9:34 AM, Suresh Marru wrote:
> > > > > Makes sense Lahiru. I will commit generated data model into commons and client stubs and server skeletons.
> > > > >
> > > > > Suresh
> > > > >
> > > > > On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> > > > >
> > > > >> Hi Suresh,
> > > > >>
> > > > >> I think to do a proper review, we need to generate code and commit to the repository. This will make life easier for developers, because everybody doesn't have to go through the hard steps to install Thrift in their machines.
> > > > >>
> > > > >> By looking in to .thrift files its difficult to navigate through the object model and hard to understand it.
> > > > >>
> > > > >> Regards
> > > > >> Lahiru
> > > > >>
> > > > >>
> > > > >> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu> wrote:
> > > > >> Hangout URL is
> > > > >> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> > > > >>
> > > > >>
> > > > >> Marlon
> > > > >>
> > > > >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> > > > >>> Hi all--
> > > > >>>
> > > > >>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> > > > >>> URL will be forthcoming.
> > > > >>>
> > > > >>>
> > > > >>> Marlon
> > > > >>>
> > > > >>
> > > > >>
> > > > >>
> > > > >> --
> > > > >> System Analyst Programmer
> > > > >> PTI Lab
> > > > >> Indiana University
> > > >
> > > >
> > > >
> > > >
> > > > --
> > > > System Analyst Programmer
> > > > PTI Lab
> > > > Indiana University
> > >
> > >
> > >
> > >
> > > --
> > > System Analyst Programmer
> > > PTI Lab
> > > Indiana University
> >
> >
> >
> >
> > --
> > Supun Kamburugamuva
> > Member, Apache Software Foundation; http://www.apache.org
> > E-mail: supun06@gmail.com; Mobile: +1 812 369 6762
> > Blog: http://supunk.blogspot.com
> >
>
>
>
>
> --
> Supun Kamburugamuva
> Member, Apache Software Foundation; http://www.apache.org
> E-mail: supun06@gmail.com; Mobile: +1 812 369 6762
> Blog: http://supunk.blogspot.com
>
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Supun Kamburugamuva <su...@gmail.com>.
Hi Suresh,
Here are couple of examples I had about using a script to generate code for
Thrift
http://git-wip-us.apache.org/repos/asf/accumulo.git
You can find the code generation script under core/src/main/scripts
Another simple project is
https://github.com/yahoo/storm-yarn
Hadoop is using prtotocol buffers and they require protocol buffers to be
installed in the system to compile hadoop. May be Hadoop can afford to do
that :)
Thanks,
Supun..
On Tue, Feb 4, 2014 at 11:36 AM, Suresh Marru <sm...@apache.org> wrote:
> Thanks Supun. Do you have any good references of public repos which you
> liked the thrift organization? I am trying to follow how cassandra is doing
> it. We also need to worry about license headers to the generated code to
> pass the release checks.
>
> Suresh
>
> On Feb 4, 2014, at 11:28 AM, Supun Kamburugamuva <su...@gmail.com>
> wrote:
>
> > I think the general practice in Thrift is to include a script that can
> do the thrift source generation if it needs to be done. The generated code
> is committed to the source repo.
> >
> > You can also use the maven plugin along with a maven profile to generate
> the code, if code generation is required.
> >
> > Thanks,
> > Supun..
> >
> >
> > On Tue, Feb 4, 2014 at 11:21 AM, Suresh Marru <sm...@apache.org> wrote:
> > Frankly we need some first hand experience and keep iterating on
> approaches on how all of this will play out across versions. Lets start
> with some approach and go from there.
> >
> > Thrift has a maven plugin in development but Chathuri confirmed even it
> will need thrift to be installed locally. It makes sense since the thrift
> generator is in C++ and is platform dependent.
> >
> > Suresh
> >
> > On Feb 4, 2014, at 11:13 AM, Lahiru Gunathilake <gl...@gmail.com>
> wrote:
> >
> > > Hi Suresh,
> > >
> > > Does thrift has a maven plugin to generate classes ? We can have
> interfaces in one module and different implementations in a separate
> module. If there's a change (API level change, the implementation modules
> should be changed, otherwise they shouldn't be compiled). But if someone
> wants the old one they can have a dependency to the old module and should
> be able to compile the impl module).
> > >
> > > Suresh, WDYT ?
> > >
> > > Regards
> > > Lahiru
> > >
> > >
> > > On Tue, Feb 4, 2014 at 11:10 AM, Suresh Marru <sm...@apache.org>
> wrote:
> > > On Feb 4, 2014, at 9:59 AM, Lahiru Gunathilake <gl...@gmail.com>
> wrote:
> > >
> > > > Hi Suresh,
> > > >
> > > > Current airavataAPI looks very clean and easy to understand. Thanks
> to your very detail comments I was able to understand the process.
> > > >
> > > > I think we need to make sure we give a simple easy to use
> experimentModel.ConfigurationData object to configure. What is our plan
> with implementing experimentModel.ConfigurationData ?
> > >
> > > Yes, we need every one's critical eye in making these configuration
> data to cover all the use cases, yet be simple and as concise as possible.
> > >
> > > For the implementation, thrift has a generator for java beans (in
> addition to standard java) which essentially makes all members to private
> with setters to return void. We can have this generated beans as a separate
> airavata-thrift-datamodel.jar and all components will use this the same way
> we now use workflow execution context or workflow tracking library. Not
> sure if this is what you asked.
> > >
> > > > Regards
> > > > Lahiru
> > > >
> > > >
> > > > On Tue, Feb 4, 2014 at 9:35 AM, Marlon Pierce <ma...@iu.edu>
> wrote:
> > > > Where do these go? In modules/thrift-interfaces?
> > >
> > > I am now poking around at other thrift based projects to find a way to
> organize these. A challenge I see if how to over-write the generated
> classes without loosing any implementations. Will post a layout for
> discussion soon.
> > >
> > > Suresh
> > >
> > >
> > > >
> > > > Marlon
> > > >
> > > > On 2/4/14 9:34 AM, Suresh Marru wrote:
> > > > > Makes sense Lahiru. I will commit generated data model into
> commons and client stubs and server skeletons.
> > > > >
> > > > > Suresh
> > > > >
> > > > > On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com>
> wrote:
> > > > >
> > > > >> Hi Suresh,
> > > > >>
> > > > >> I think to do a proper review, we need to generate code and
> commit to the repository. This will make life easier for developers,
> because everybody doesn't have to go through the hard steps to install
> Thrift in their machines.
> > > > >>
> > > > >> By looking in to .thrift files its difficult to navigate through
> the object model and hard to understand it.
> > > > >>
> > > > >> Regards
> > > > >> Lahiru
> > > > >>
> > > > >>
> > > > >> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu>
> wrote:
> > > > >> Hangout URL is
> > > > >>
> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> > > > >>
> > > > >>
> > > > >> Marlon
> > > > >>
> > > > >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> > > > >>> Hi all--
> > > > >>>
> > > > >>> We'll have a hangout at 1:00 pm US EST to discuss the subject.
> Hangout
> > > > >>> URL will be forthcoming.
> > > > >>>
> > > > >>>
> > > > >>> Marlon
> > > > >>>
> > > > >>
> > > > >>
> > > > >>
> > > > >> --
> > > > >> System Analyst Programmer
> > > > >> PTI Lab
> > > > >> Indiana University
> > > >
> > > >
> > > >
> > > >
> > > > --
> > > > System Analyst Programmer
> > > > PTI Lab
> > > > Indiana University
> > >
> > >
> > >
> > >
> > > --
> > > System Analyst Programmer
> > > PTI Lab
> > > Indiana University
> >
> >
> >
> >
> > --
> > Supun Kamburugamuva
> > Member, Apache Software Foundation; http://www.apache.org
> > E-mail: supun06@gmail.com; Mobile: +1 812 369 6762
> > Blog: http://supunk.blogspot.com
> >
>
>
--
Supun Kamburugamuva
Member, Apache Software Foundation; http://www.apache.org
E-mail: supun06@gmail.com; Mobile: +1 812 369 6762
Blog: http://supunk.blogspot.com
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Suresh Marru <sm...@apache.org>.
Thanks Supun. Do you have any good references of public repos which you liked the thrift organization? I am trying to follow how cassandra is doing it. We also need to worry about license headers to the generated code to pass the release checks.
Suresh
On Feb 4, 2014, at 11:28 AM, Supun Kamburugamuva <su...@gmail.com> wrote:
> I think the general practice in Thrift is to include a script that can do the thrift source generation if it needs to be done. The generated code is committed to the source repo.
>
> You can also use the maven plugin along with a maven profile to generate the code, if code generation is required.
>
> Thanks,
> Supun..
>
>
> On Tue, Feb 4, 2014 at 11:21 AM, Suresh Marru <sm...@apache.org> wrote:
> Frankly we need some first hand experience and keep iterating on approaches on how all of this will play out across versions. Lets start with some approach and go from there.
>
> Thrift has a maven plugin in development but Chathuri confirmed even it will need thrift to be installed locally. It makes sense since the thrift generator is in C++ and is platform dependent.
>
> Suresh
>
> On Feb 4, 2014, at 11:13 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
>
> > Hi Suresh,
> >
> > Does thrift has a maven plugin to generate classes ? We can have interfaces in one module and different implementations in a separate module. If there's a change (API level change, the implementation modules should be changed, otherwise they shouldn't be compiled). But if someone wants the old one they can have a dependency to the old module and should be able to compile the impl module).
> >
> > Suresh, WDYT ?
> >
> > Regards
> > Lahiru
> >
> >
> > On Tue, Feb 4, 2014 at 11:10 AM, Suresh Marru <sm...@apache.org> wrote:
> > On Feb 4, 2014, at 9:59 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> >
> > > Hi Suresh,
> > >
> > > Current airavataAPI looks very clean and easy to understand. Thanks to your very detail comments I was able to understand the process.
> > >
> > > I think we need to make sure we give a simple easy to use experimentModel.ConfigurationData object to configure. What is our plan with implementing experimentModel.ConfigurationData ?
> >
> > Yes, we need every one’s critical eye in making these configuration data to cover all the use cases, yet be simple and as concise as possible.
> >
> > For the implementation, thrift has a generator for java beans (in addition to standard java) which essentially makes all members to private with setters to return void. We can have this generated beans as a separate airavata-thrift-datamodel.jar and all components will use this the same way we now use workflow execution context or workflow tracking library. Not sure if this is what you asked.
> >
> > > Regards
> > > Lahiru
> > >
> > >
> > > On Tue, Feb 4, 2014 at 9:35 AM, Marlon Pierce <ma...@iu.edu> wrote:
> > > Where do these go? In modules/thrift-interfaces?
> >
> > I am now poking around at other thrift based projects to find a way to organize these. A challenge I see if how to over-write the generated classes without loosing any implementations. Will post a layout for discussion soon.
> >
> > Suresh
> >
> >
> > >
> > > Marlon
> > >
> > > On 2/4/14 9:34 AM, Suresh Marru wrote:
> > > > Makes sense Lahiru. I will commit generated data model into commons and client stubs and server skeletons.
> > > >
> > > > Suresh
> > > >
> > > > On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> > > >
> > > >> Hi Suresh,
> > > >>
> > > >> I think to do a proper review, we need to generate code and commit to the repository. This will make life easier for developers, because everybody doesn't have to go through the hard steps to install Thrift in their machines.
> > > >>
> > > >> By looking in to .thrift files its difficult to navigate through the object model and hard to understand it.
> > > >>
> > > >> Regards
> > > >> Lahiru
> > > >>
> > > >>
> > > >> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu> wrote:
> > > >> Hangout URL is
> > > >> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> > > >>
> > > >>
> > > >> Marlon
> > > >>
> > > >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> > > >>> Hi all--
> > > >>>
> > > >>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> > > >>> URL will be forthcoming.
> > > >>>
> > > >>>
> > > >>> Marlon
> > > >>>
> > > >>
> > > >>
> > > >>
> > > >> --
> > > >> System Analyst Programmer
> > > >> PTI Lab
> > > >> Indiana University
> > >
> > >
> > >
> > >
> > > --
> > > System Analyst Programmer
> > > PTI Lab
> > > Indiana University
> >
> >
> >
> >
> > --
> > System Analyst Programmer
> > PTI Lab
> > Indiana University
>
>
>
>
> --
> Supun Kamburugamuva
> Member, Apache Software Foundation; http://www.apache.org
> E-mail: supun06@gmail.com; Mobile: +1 812 369 6762
> Blog: http://supunk.blogspot.com
>
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Supun Kamburugamuva <su...@gmail.com>.
I think the general practice in Thrift is to include a script that can do
the thrift source generation if it needs to be done. The generated code is
committed to the source repo.
You can also use the maven plugin along with a maven profile to generate
the code, if code generation is required.
Thanks,
Supun..
On Tue, Feb 4, 2014 at 11:21 AM, Suresh Marru <sm...@apache.org> wrote:
> Frankly we need some first hand experience and keep iterating on
> approaches on how all of this will play out across versions. Lets start
> with some approach and go from there.
>
> Thrift has a maven plugin in development but Chathuri confirmed even it
> will need thrift to be installed locally. It makes sense since the thrift
> generator is in C++ and is platform dependent.
>
> Suresh
>
> On Feb 4, 2014, at 11:13 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
>
> > Hi Suresh,
> >
> > Does thrift has a maven plugin to generate classes ? We can have
> interfaces in one module and different implementations in a separate
> module. If there's a change (API level change, the implementation modules
> should be changed, otherwise they shouldn't be compiled). But if someone
> wants the old one they can have a dependency to the old module and should
> be able to compile the impl module).
> >
> > Suresh, WDYT ?
> >
> > Regards
> > Lahiru
> >
> >
> > On Tue, Feb 4, 2014 at 11:10 AM, Suresh Marru <sm...@apache.org> wrote:
> > On Feb 4, 2014, at 9:59 AM, Lahiru Gunathilake <gl...@gmail.com>
> wrote:
> >
> > > Hi Suresh,
> > >
> > > Current airavataAPI looks very clean and easy to understand. Thanks to
> your very detail comments I was able to understand the process.
> > >
> > > I think we need to make sure we give a simple easy to use
> experimentModel.ConfigurationData object to configure. What is our plan
> with implementing experimentModel.ConfigurationData ?
> >
> > Yes, we need every one's critical eye in making these configuration data
> to cover all the use cases, yet be simple and as concise as possible.
> >
> > For the implementation, thrift has a generator for java beans (in
> addition to standard java) which essentially makes all members to private
> with setters to return void. We can have this generated beans as a separate
> airavata-thrift-datamodel.jar and all components will use this the same way
> we now use workflow execution context or workflow tracking library. Not
> sure if this is what you asked.
> >
> > > Regards
> > > Lahiru
> > >
> > >
> > > On Tue, Feb 4, 2014 at 9:35 AM, Marlon Pierce <ma...@iu.edu> wrote:
> > > Where do these go? In modules/thrift-interfaces?
> >
> > I am now poking around at other thrift based projects to find a way to
> organize these. A challenge I see if how to over-write the generated
> classes without loosing any implementations. Will post a layout for
> discussion soon.
> >
> > Suresh
> >
> >
> > >
> > > Marlon
> > >
> > > On 2/4/14 9:34 AM, Suresh Marru wrote:
> > > > Makes sense Lahiru. I will commit generated data model into commons
> and client stubs and server skeletons.
> > > >
> > > > Suresh
> > > >
> > > > On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com>
> wrote:
> > > >
> > > >> Hi Suresh,
> > > >>
> > > >> I think to do a proper review, we need to generate code and commit
> to the repository. This will make life easier for developers, because
> everybody doesn't have to go through the hard steps to install Thrift in
> their machines.
> > > >>
> > > >> By looking in to .thrift files its difficult to navigate through
> the object model and hard to understand it.
> > > >>
> > > >> Regards
> > > >> Lahiru
> > > >>
> > > >>
> > > >> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu>
> wrote:
> > > >> Hangout URL is
> > > >> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> > > >>
> > > >>
> > > >> Marlon
> > > >>
> > > >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> > > >>> Hi all--
> > > >>>
> > > >>> We'll have a hangout at 1:00 pm US EST to discuss the subject.
> Hangout
> > > >>> URL will be forthcoming.
> > > >>>
> > > >>>
> > > >>> Marlon
> > > >>>
> > > >>
> > > >>
> > > >>
> > > >> --
> > > >> System Analyst Programmer
> > > >> PTI Lab
> > > >> Indiana University
> > >
> > >
> > >
> > >
> > > --
> > > System Analyst Programmer
> > > PTI Lab
> > > Indiana University
> >
> >
> >
> >
> > --
> > System Analyst Programmer
> > PTI Lab
> > Indiana University
>
>
--
Supun Kamburugamuva
Member, Apache Software Foundation; http://www.apache.org
E-mail: supun06@gmail.com; Mobile: +1 812 369 6762
Blog: http://supunk.blogspot.com
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Suresh Marru <sm...@apache.org>.
Frankly we need some first hand experience and keep iterating on approaches on how all of this will play out across versions. Lets start with some approach and go from there.
Thrift has a maven plugin in development but Chathuri confirmed even it will need thrift to be installed locally. It makes sense since the thrift generator is in C++ and is platform dependent.
Suresh
On Feb 4, 2014, at 11:13 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> Hi Suresh,
>
> Does thrift has a maven plugin to generate classes ? We can have interfaces in one module and different implementations in a separate module. If there's a change (API level change, the implementation modules should be changed, otherwise they shouldn't be compiled). But if someone wants the old one they can have a dependency to the old module and should be able to compile the impl module).
>
> Suresh, WDYT ?
>
> Regards
> Lahiru
>
>
> On Tue, Feb 4, 2014 at 11:10 AM, Suresh Marru <sm...@apache.org> wrote:
> On Feb 4, 2014, at 9:59 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
>
> > Hi Suresh,
> >
> > Current airavataAPI looks very clean and easy to understand. Thanks to your very detail comments I was able to understand the process.
> >
> > I think we need to make sure we give a simple easy to use experimentModel.ConfigurationData object to configure. What is our plan with implementing experimentModel.ConfigurationData ?
>
> Yes, we need every one’s critical eye in making these configuration data to cover all the use cases, yet be simple and as concise as possible.
>
> For the implementation, thrift has a generator for java beans (in addition to standard java) which essentially makes all members to private with setters to return void. We can have this generated beans as a separate airavata-thrift-datamodel.jar and all components will use this the same way we now use workflow execution context or workflow tracking library. Not sure if this is what you asked.
>
> > Regards
> > Lahiru
> >
> >
> > On Tue, Feb 4, 2014 at 9:35 AM, Marlon Pierce <ma...@iu.edu> wrote:
> > Where do these go? In modules/thrift-interfaces?
>
> I am now poking around at other thrift based projects to find a way to organize these. A challenge I see if how to over-write the generated classes without loosing any implementations. Will post a layout for discussion soon.
>
> Suresh
>
>
> >
> > Marlon
> >
> > On 2/4/14 9:34 AM, Suresh Marru wrote:
> > > Makes sense Lahiru. I will commit generated data model into commons and client stubs and server skeletons.
> > >
> > > Suresh
> > >
> > > On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> > >
> > >> Hi Suresh,
> > >>
> > >> I think to do a proper review, we need to generate code and commit to the repository. This will make life easier for developers, because everybody doesn't have to go through the hard steps to install Thrift in their machines.
> > >>
> > >> By looking in to .thrift files its difficult to navigate through the object model and hard to understand it.
> > >>
> > >> Regards
> > >> Lahiru
> > >>
> > >>
> > >> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu> wrote:
> > >> Hangout URL is
> > >> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> > >>
> > >>
> > >> Marlon
> > >>
> > >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> > >>> Hi all--
> > >>>
> > >>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> > >>> URL will be forthcoming.
> > >>>
> > >>>
> > >>> Marlon
> > >>>
> > >>
> > >>
> > >>
> > >> --
> > >> System Analyst Programmer
> > >> PTI Lab
> > >> Indiana University
> >
> >
> >
> >
> > --
> > System Analyst Programmer
> > PTI Lab
> > Indiana University
>
>
>
>
> --
> System Analyst Programmer
> PTI Lab
> Indiana University
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Lahiru Gunathilake <gl...@gmail.com>.
Hi Suresh,
Does thrift has a maven plugin to generate classes ? We can have interfaces
in one module and different implementations in a separate module. If
there's a change (API level change, the implementation modules should be
changed, otherwise they shouldn't be compiled). But if someone wants the
old one they can have a dependency to the old module and should be able to
compile the impl module).
Suresh, WDYT ?
Regards
Lahiru
On Tue, Feb 4, 2014 at 11:10 AM, Suresh Marru <sm...@apache.org> wrote:
> On Feb 4, 2014, at 9:59 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
>
> > Hi Suresh,
> >
> > Current airavataAPI looks very clean and easy to understand. Thanks to
> your very detail comments I was able to understand the process.
> >
> > I think we need to make sure we give a simple easy to use
> experimentModel.ConfigurationData object to configure. What is our plan
> with implementing experimentModel.ConfigurationData ?
>
> Yes, we need every one's critical eye in making these configuration data
> to cover all the use cases, yet be simple and as concise as possible.
>
> For the implementation, thrift has a generator for java beans (in addition
> to standard java) which essentially makes all members to private with
> setters to return void. We can have this generated beans as a separate
> airavata-thrift-datamodel.jar and all components will use this the same way
> we now use workflow execution context or workflow tracking library. Not
> sure if this is what you asked.
>
> > Regards
> > Lahiru
> >
> >
> > On Tue, Feb 4, 2014 at 9:35 AM, Marlon Pierce <ma...@iu.edu> wrote:
> > Where do these go? In modules/thrift-interfaces?
>
> I am now poking around at other thrift based projects to find a way to
> organize these. A challenge I see if how to over-write the generated
> classes without loosing any implementations. Will post a layout for
> discussion soon.
>
> Suresh
>
>
> >
> > Marlon
> >
> > On 2/4/14 9:34 AM, Suresh Marru wrote:
> > > Makes sense Lahiru. I will commit generated data model into commons
> and client stubs and server skeletons.
> > >
> > > Suresh
> > >
> > > On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com>
> wrote:
> > >
> > >> Hi Suresh,
> > >>
> > >> I think to do a proper review, we need to generate code and commit to
> the repository. This will make life easier for developers, because
> everybody doesn't have to go through the hard steps to install Thrift in
> their machines.
> > >>
> > >> By looking in to .thrift files its difficult to navigate through the
> object model and hard to understand it.
> > >>
> > >> Regards
> > >> Lahiru
> > >>
> > >>
> > >> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu>
> wrote:
> > >> Hangout URL is
> > >> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> > >>
> > >>
> > >> Marlon
> > >>
> > >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> > >>> Hi all--
> > >>>
> > >>> We'll have a hangout at 1:00 pm US EST to discuss the subject.
> Hangout
> > >>> URL will be forthcoming.
> > >>>
> > >>>
> > >>> Marlon
> > >>>
> > >>
> > >>
> > >>
> > >> --
> > >> System Analyst Programmer
> > >> PTI Lab
> > >> Indiana University
> >
> >
> >
> >
> > --
> > System Analyst Programmer
> > PTI Lab
> > Indiana University
>
>
--
System Analyst Programmer
PTI Lab
Indiana University
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Suresh Marru <sm...@apache.org>.
On Feb 4, 2014, at 9:59 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> Hi Suresh,
>
> Current airavataAPI looks very clean and easy to understand. Thanks to your very detail comments I was able to understand the process.
>
> I think we need to make sure we give a simple easy to use experimentModel.ConfigurationData object to configure. What is our plan with implementing experimentModel.ConfigurationData ?
Yes, we need every one’s critical eye in making these configuration data to cover all the use cases, yet be simple and as concise as possible.
For the implementation, thrift has a generator for java beans (in addition to standard java) which essentially makes all members to private with setters to return void. We can have this generated beans as a separate airavata-thrift-datamodel.jar and all components will use this the same way we now use workflow execution context or workflow tracking library. Not sure if this is what you asked.
> Regards
> Lahiru
>
>
> On Tue, Feb 4, 2014 at 9:35 AM, Marlon Pierce <ma...@iu.edu> wrote:
> Where do these go? In modules/thrift-interfaces?
I am now poking around at other thrift based projects to find a way to organize these. A challenge I see if how to over-write the generated classes without loosing any implementations. Will post a layout for discussion soon.
Suresh
>
> Marlon
>
> On 2/4/14 9:34 AM, Suresh Marru wrote:
> > Makes sense Lahiru. I will commit generated data model into commons and client stubs and server skeletons.
> >
> > Suresh
> >
> > On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> >
> >> Hi Suresh,
> >>
> >> I think to do a proper review, we need to generate code and commit to the repository. This will make life easier for developers, because everybody doesn't have to go through the hard steps to install Thrift in their machines.
> >>
> >> By looking in to .thrift files its difficult to navigate through the object model and hard to understand it.
> >>
> >> Regards
> >> Lahiru
> >>
> >>
> >> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu> wrote:
> >> Hangout URL is
> >> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> >>
> >>
> >> Marlon
> >>
> >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> >>> Hi all--
> >>>
> >>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> >>> URL will be forthcoming.
> >>>
> >>>
> >>> Marlon
> >>>
> >>
> >>
> >>
> >> --
> >> System Analyst Programmer
> >> PTI Lab
> >> Indiana University
>
>
>
>
> --
> System Analyst Programmer
> PTI Lab
> Indiana University
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Lahiru Gunathilake <gl...@gmail.com>.
Hi Suresh,
Current airavataAPI looks very clean and easy to understand. Thanks to your
very detail comments I was able to understand the process.
I think we need to make sure we give a simple easy to
use experimentModel.ConfigurationData object to configure. What is our plan
with implementing experimentModel.ConfigurationData ?
Regards
Lahiru
On Tue, Feb 4, 2014 at 9:35 AM, Marlon Pierce <ma...@iu.edu> wrote:
> Where do these go? In modules/thrift-interfaces?
>
> Marlon
>
> On 2/4/14 9:34 AM, Suresh Marru wrote:
> > Makes sense Lahiru. I will commit generated data model into commons and
> client stubs and server skeletons.
> >
> > Suresh
> >
> > On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com>
> wrote:
> >
> >> Hi Suresh,
> >>
> >> I think to do a proper review, we need to generate code and commit to
> the repository. This will make life easier for developers, because
> everybody doesn't have to go through the hard steps to install Thrift in
> their machines.
> >>
> >> By looking in to .thrift files its difficult to navigate through the
> object model and hard to understand it.
> >>
> >> Regards
> >> Lahiru
> >>
> >>
> >> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu> wrote:
> >> Hangout URL is
> >> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
> >>
> >>
> >> Marlon
> >>
> >> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> >>> Hi all--
> >>>
> >>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> >>> URL will be forthcoming.
> >>>
> >>>
> >>> Marlon
> >>>
> >>
> >>
> >>
> >> --
> >> System Analyst Programmer
> >> PTI Lab
> >> Indiana University
>
>
--
System Analyst Programmer
PTI Lab
Indiana University
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Marlon Pierce <ma...@iu.edu>.
Where do these go? In modules/thrift-interfaces?
Marlon
On 2/4/14 9:34 AM, Suresh Marru wrote:
> Makes sense Lahiru. I will commit generated data model into commons and client stubs and server skeletons.
>
> Suresh
>
> On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
>
>> Hi Suresh,
>>
>> I think to do a proper review, we need to generate code and commit to the repository. This will make life easier for developers, because everybody doesn't have to go through the hard steps to install Thrift in their machines.
>>
>> By looking in to .thrift files its difficult to navigate through the object model and hard to understand it.
>>
>> Regards
>> Lahiru
>>
>>
>> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu> wrote:
>> Hangout URL is
>> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
>>
>>
>> Marlon
>>
>> On 2/3/14 9:14 AM, Marlon Pierce wrote:
>>> Hi all--
>>>
>>> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
>>> URL will be forthcoming.
>>>
>>>
>>> Marlon
>>>
>>
>>
>>
>> --
>> System Analyst Programmer
>> PTI Lab
>> Indiana University
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Suresh Marru <sm...@apache.org>.
Makes sense Lahiru. I will commit generated data model into commons and client stubs and server skeletons.
Suresh
On Feb 4, 2014, at 9:18 AM, Lahiru Gunathilake <gl...@gmail.com> wrote:
> Hi Suresh,
>
> I think to do a proper review, we need to generate code and commit to the repository. This will make life easier for developers, because everybody doesn't have to go through the hard steps to install Thrift in their machines.
>
> By looking in to .thrift files its difficult to navigate through the object model and hard to understand it.
>
> Regards
> Lahiru
>
>
> On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu> wrote:
> Hangout URL is
> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
>
>
> Marlon
>
> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> > Hi all--
> >
> > We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> > URL will be forthcoming.
> >
> >
> > Marlon
> >
>
>
>
>
> --
> System Analyst Programmer
> PTI Lab
> Indiana University
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Lahiru Gunathilake <gl...@gmail.com>.
Hi Suresh,
I think to do a proper review, we need to generate code and commit to the
repository. This will make life easier for developers, because everybody
doesn't have to go through the hard steps to install Thrift in their
machines.
By looking in to .thrift files its difficult to navigate through the object
model and hard to understand it.
Regards
Lahiru
On Mon, Feb 3, 2014 at 1:04 PM, Marlon Pierce <ma...@iu.edu> wrote:
> Hangout URL is
> https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
>
>
> Marlon
>
> On 2/3/14 9:14 AM, Marlon Pierce wrote:
> > Hi all--
> >
> > We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> > URL will be forthcoming.
> >
> >
> > Marlon
> >
>
>
--
System Analyst Programmer
PTI Lab
Indiana University
Re: Hangout: Orchestrator, Registry, GFAC, and API recap and plans
Posted by Marlon Pierce <ma...@iu.edu>.
Hangout URL is
https://plus.google.com/hangouts/_/72cpjvt6sl94fnn4hs10bibi6g?hl=en
Marlon
On 2/3/14 9:14 AM, Marlon Pierce wrote:
> Hi all--
>
> We'll have a hangout at 1:00 pm US EST to discuss the subject. Hangout
> URL will be forthcoming.
>
>
> Marlon
>