On Thu, Mar 15, 2012 at 12:20, Vries, Theo J.A. de [imotec bv]
<t [dot] j [dot] a [dot] devries [..] ...> wrote:
> 2012/3/12 Sylvain Joyeux <sylvain [dot] joyeux [..] ...>
>>
>> I recently wrote some documentation about a few "patterns" in data
>> processing (patterns that has some specialized support in Rock). A lot
>> of the text deals with explaining the principles, which is orocos- and
>> rock- unspecific, so I thought I would query some feedback from the
>> community
>>
>> The documentation is here:
>>
>>   http://rock-robotics.org/documentation/data_processing/index.html
>>
>> Comments are very welcome !
>>
>> Sylvain
>> --
>>
>> Orocos-Dev mailing list
>> Orocos-Dev [..] ...
>> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-dev
>
>
> I like it very much. Have written something alike myself, but this is much
> more elaborate and thorough.

+1. I also like it very much, but I have not written something similar (yet).

In my work I am trying to define these kinds of "patterns" more
formally, i.e. in terms of data
and control flows. I use Petri nets (with extensions like colors,
hierarchy, priorities and timing)
for this purpose. The definition is strict and precise, but makes the
patterns somehow difficult
to communicate to human readers...

However, with a formal definitions it is easier to show how different
patterns can be combined
and analyze how they work together. Collection of patterns can be then
then formalized as DSL
and from that point one can apply all the MDE to work with them (e.g.
build editors, validate the
model, generate code, etc.).

>
> I also would like to refer to it; can you maybe add a bibtex entry that says
> how you want it to be referred to?
>
> Minor remarks:
> -
> pg http://rock-robotics.org/documentation/data_processing/timestamping.html typo QNX
> \emph{or} has

For indoor control applications you can use PTP instead of NTP for
clock synchronization .
Especially in wired networks (when the bandwidth should not be a
problem) it converge
faster and also gives you better accuracy.

> -
> pg http://rock-robotics.org/documentation/data_processing/stream_aligner.html missing
> word(s)
> The position of these sensors w.r.t. the body is therefore
> -
> pg http://rock-robotics.org/documentation/data_processing/stream_aligner.html and http://rock-robotics.org/documentation/data_processing/stream_aligner_runtime.html that
> -> than
> earlier timestamp that the last sample
>
> cheers, Theo
>
>
> --
> Orocos-Users mailing list
> Orocos-Users [..] ...
> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-users
>

On Thu, Mar 15, 2012 at 12:20, Vries, Theo J.A. de [imotec bv]
<t [dot] j [dot] a [dot] devries [..] ...> wrote:
> 2012/3/12 Sylvain Joyeux <sylvain [dot] joyeux [..] ...>
>>
>> I recently wrote some documentation about a few "patterns" in data
>> processing (patterns that has some specialized support in Rock). A lot
>> of the text deals with explaining the principles, which is orocos- and
>> rock- unspecific, so I thought I would query some feedback from the
>> community
>>
>> The documentation is here:
>>
>>   http://rock-robotics.org/documentation/data_processing/index.html
>>
>> Comments are very welcome !
>>
>> Sylvain
>> --
>>
>> Orocos-Dev mailing list
>> Orocos-Dev [..] ...
>> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-dev
>
>
> I like it very much. Have written something alike myself, but this is much
> more elaborate and thorough.

+1. I also like it very much, but I have not written something similar (yet).

In my work I am trying to define these kinds of "patterns" more
formally, i.e. in terms of data
and control flows. I use Petri nets (with extensions like colors,
hierarchy, priorities and timing)
for this purpose. The definition is strict and precise, but makes the
patterns somehow difficult
to communicate to human readers...

However, with a formal definitions it is easier to show how different
patterns can be combined
and analyze how they work together. Collection of patterns can be then
then formalized as DSL
and from that point one can apply all the MDE to work with them (e.g.
build editors, validate the
model, generate code, etc.).

>
> I also would like to refer to it; can you maybe add a bibtex entry that says
> how you want it to be referred to?
>
> Minor remarks:
> -
> pg http://rock-robotics.org/documentation/data_processing/timestamping.html typo QNX
> \emph{or} has

For indoor control applications you can use PTP instead of NTP for
clock synchronization .
Especially in wired networks (when the bandwidth should not be a
problem) it converge
faster and also gives you better accuracy.

> -
> pg http://rock-robotics.org/documentation/data_processing/stream_aligner.html missing
> word(s)
> The position of these sensors w.r.t. the body is therefore
> -
> pg http://rock-robotics.org/documentation/data_processing/stream_aligner.html and http://rock-robotics.org/documentation/data_processing/stream_aligner_runtime.html that
> -> than
> earlier timestamp that the last sample
>
> cheers, Theo
>
>
> --
> Orocos-Users mailing list
> Orocos-Users [..] ...
> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-users
>

Nice work Sylvain.

Do you mind if I prsent some of the concepts in the BRICS Wikibook (with a
reference to your work, of course!!)

Davide

On Tue, Mar 13, 2012 at 9:54 AM, Peter Soetens <peter [..] ...>wrote:

> On Mon, Mar 12, 2012 at 5:14 PM, Sylvain Joyeux <sylvain [dot] joyeux [..] ...>wrote:
>
>> I recently wrote some documentation about a few "patterns" in data
>> processing (patterns that has some specialized support in Rock). A lot
>> of the text deals with explaining the principles, which is orocos- and
>> rock- unspecific, so I thought I would query some feedback from the
>> community
>>
>> The documentation is here:
>>
>> http://rock-robotics.org/documentation/data_processing/index.html
>>
>> Comments are very welcome !
>>
>
> Nice contribution and generic solution ! A definite must for every system
> exceeding the trivial.
>
> Peter
>
>
>
> --
> Orocos-Dev mailing list
> Orocos-Dev [..] ...
> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-dev
>
>

Nice work Sylvain.

Do you mind if I prsent some of the concepts in the BRICS Wikibook (with a
reference to your work, of course!!)

Davide

On Tue, Mar 13, 2012 at 9:54 AM, Peter Soetens <peter [..] ...>wrote:

> On Mon, Mar 12, 2012 at 5:14 PM, Sylvain Joyeux <sylvain [dot] joyeux [..] ...>wrote:
>
>> I recently wrote some documentation about a few "patterns" in data
>> processing (patterns that has some specialized support in Rock). A lot
>> of the text deals with explaining the principles, which is orocos- and
>> rock- unspecific, so I thought I would query some feedback from the
>> community
>>
>> The documentation is here:
>>
>> http://rock-robotics.org/documentation/data_processing/index.html
>>
>> Comments are very welcome !
>>
>
> Nice contribution and generic solution ! A definite must for every system
> exceeding the trivial.
>
> Peter
>
>
>
> --
> Orocos-Dev mailing list
> Orocos-Dev [..] ...
> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-dev
>
>

On 03/13/2012 01:53 PM, Davide Faconti wrote:
> Nice work Sylvain.
>
> Do you mind if I prsent some of the concepts in the BRICS Wikibook (with
> a reference to your work, of course!!)

As long as Rock and the rock website are explicitly mentioned, I have no
problem at all.

You can find figures in svg this git repository:

http://gitorious.org/rock/doc

Sylvain