[openrtm-users 01485] Re: regarding rtprint and rtinject

[Geoffrey Biggs]

Hi Tony,

Thanks for the feeback.

On 29/10/10 07:53, Tony Kuo wrote:
> Hi Geoff:
> The first thing that came to my mind are
> to improve the speed (which you already know) and also to allow
> connections to be restored.
> 
> The reason I don't find rts-shells useful is that it doesn't restore the
> connections between the components (correct me if I am not wrong).

I assume you're referring to rtresurrect here.

> And restoring the connections is probably the most time consuming
> process in developing my system.
> Because I can do start, stop, activate and deactive in the Eclipse-based
> RTSystemEditor.
> 
> While testing 1 particular component in a large system, I sometime find
> it necessary to start, stop and reconnect the connections around just
> that component.
> Running my bash script for the whole system would duplicate some of the
> connections that are already there (there is no way to find out whether
> they are connected already or not)
> 
> If somehow, the scripts allow some kind of check on whether the
> component is already alive, the connection is already done or the
> components are all in Activated state (not in error or inactivated state)
> it would be quite convenient I think.

I've just committed a change to trunk that makes rtresurrect and
rtteardown use connector IDs. This means that rtresurrect will now skip
connections that already exist. You can use it to restore a network to
its previous state after removing only some of the connections, without
any existing connections being duplicated.

> Another small thing is if an installation has pre-requisite to run, for
> example, rtcshell requires rtctree, there should be a dependency check
> during installation to make it easier for users?

Unfortunately, the Python distutils don't yet provide a way to easily
check that pre-requisites have been installed. I might try hacking
something in.

Geoff