aperiodic component 'as fast as possible'

Submitted by gah on Thu, 2012-04-26 09:18

All,

The documentation describes the use of 'trigger()' in an aperiodic component to get its updateHook() called again (in absence of any other sources of 'triggers').

Am I correct in assuming that if there is no blocking read, or sleep, or something similar in the updateHook(), the execution engine will basically keep calling the updateHook of a component as fast as possible? All depending on other components and pending commands and events I think?

A test component does indeed start using 100% cpu time, which seems to suggest my assumption is correct.

‹ FBSched: segfault at deployer exit Error installing orocos-toolchain ›
»

aperiodic component 'as fast as possible'

Submitted by Steven Bellens on Thu, 2012-04-26 11:00.

2012/4/26 g ah <gaohml [..] ...>:
>
> All,
>
> The documentation describes the use of 'trigger()' in an aperiodic component to get its updateHook() called again (in absence of any other sources of 'triggers').
>
> Am I correct in assuming that if there is no blocking read, or sleep, or something similar in the updateHook(), the execution engine will basically keep calling the updateHook of a component as fast as possible? All depending on other components and pending commands and events I think?
>
> A test component does indeed start using 100% cpu time, which seems to suggest my assumption is correct.

Yes, to execute the updateHook as fast as possible, make your
component aperiodic and, at the end of updateHook(), call trigger().

Steven

>
>
>
> --
> Orocos-Users mailing list
> Orocos-Users [..] ...
> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-users
>

»

aperiodic component 'as fast as possible'

Submitted by gah on Thu, 2012-04-26 16:36.

Steven Bellens wrote:
> 2012/4/26 g ah <gaohml [..] ...>:
>> All,
>>
>> The documentation describes the use of 'trigger()' in an aperiodic component to get its updateHook() called again (in absence of any other sources of 'triggers').
>>
>> Am I correct in assuming that if there is no blocking read, or sleep, or something similar in the updateHook(), the execution engine will basically keep calling the updateHook of a component as fast as possible? All depending on other components and pending commands and events I think?
>>
>> A test component does indeed start using 100% cpu time, which seems to suggest my assumption is correct.
>
> Yes, to execute the updateHook as fast as possible, make your
> component aperiodic and, at the end of updateHook(), call trigger().

Thanks for the confirmation.

»