[Rock-dev] place for general_processing packages

Javier Hidalgo Carrió javier.hidalgo_carrio at dfki.de
Fri Feb 1 11:04:19 CET 2013


On 02/01/2013 09:22 AM, Felix Rehrmann wrote:
> Hi all,
>
> just wanted to restart this short discussion on the general_processing
> packages with first bringing forward a proposal for a new package
> catogery (folder) for data processing stuff that does not care about the
> meaning of the data or only supports data processing? Initial that could be:
>    - transformer,
>    - aggreagator,
>    - general_processing,
>    - are there other packages coming to your mind?
>
> I think, there will be more in the future. It could be place for a whole
> bunch of
> time series processing algorithms, filters and generic learning
> algorithms (although the category learning could be used for latter).
>
> That category could be called:
> - data
> - data_processing
> - process(ing)
> - data_tools
> - ...
>
> I would put the general_processing stuff in there and propose the
> following names:
> - type_to_vector for the library
> - type_to_vector or timeseries for the orogen package
>
> Would be nice to here your opinion.
+1
I don't see the reason of putting /transformer/ and /aggregator/ in 
/drivers//
/data_processing/ could be one good possibility as there is 
/image_processing/

Other possibility is into /base/ like /base/processing//
but I don't see any example where there is library and orogen packages 
organization in subfolders of another library and orogen packages 
because in this way /image_processing/ could be inside /data_processing./

Javier.

>
> Felix
>
>
> Am 04.12.2012 14:08, schrieb Alexander Duda:
>> On 12/04/2012 01:28 PM, Jakob Schwendner wrote:
>>> On 12/04/2012 12:51 PM, Felix Rehrmann wrote:
>>>> Hi all,
>>>>
>>>> currently I'm writing a component that provides the ability to add ports
>>>> dynamically.
>>>> The data from these port will automatically be converted into doubles
>>>> and collected into one or more vectors, which are buffered and made
>>>> available inside the component. The purpose of that is to integrate
>>>> generic algorithms (e.g. learning algorithm) that only take vector as
>>>> input. So the same algorithm can be used for
>>>> a different set of input data.
>>>>
>>>> There are two packages general_processing and orogen/general_processing,
>>>> which are currently developed locally. I want to put them into the
>>>> public rock part. First best choice would be to put them into tools. But
>>>> as there is already the orogen-package the standard
>>>> library/component-naming would not work.
>>>>
>>>> So any suggestions, where to put the packages? How to solve this?
>>>>
>>>> Thanks and all the best!
>>>> Felix
>>>>
>>> I would put that en par with the aggregator/transformer. Those are in
>>> drivers.
>>>
>>> Jakob
>> +1
>>
>> I am not so happy about the name "general_processing" because this can
>> basically mean anything. Therefore, I would really look for a more
>> descriptive name.
>>
>> Maybe vector_processing or sample_demux would work better for this kind
>> of base orogon component.
>>
>> Alex
>>
>


-- 
Javier Hidalgo Carrió
ESA - NPI Programme
Researcher

DFKI Bremen
Robotics Innovation Center
Robert-Hooke-Straße 5
28359 Bremen, Germany
http://robotik.dfki-bremen.de

Phone:+49(0)421 17845 6661
Fax: +49(0)421 17845 4150

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.dfki.de/pipermail/rock-dev/attachments/20130201/41e508b0/attachment.htm 


More information about the Rock-dev mailing list