You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently it is only possible to use the labgrid-client tool with the distributed infrastructure - deployed exporter, coordinator, and so on.
When trying to use labgrid-client with only a local environment configuration where no remote place is specified and no coordinator address is supplied, the client errors out with message:
No RemotePlace found in configuration file
IMHO it would be nice to close the feature gap of the client for local usage by either use the locally defined resources directly for some of it's commands like io, or console, or start coordinator&exporter on-demand internally in some way.
The text was updated successfully, but these errors were encountered:
Hi, I've been experimenting to get labgrid-client to work without the coordinator deployed. I'd like to get some opinions of if this design could be useful.
You can see a rough implementation in this commit. The diff looks large, but there aren't actually that many meaningful changes.
I've split the client into a remote and local part. The remote part works as the client does now. The local part can interact with devices, but can't do anything with places.
The existing ClientSession class has been split into a base ClientSession and subclasses of RemoteSession and LocalSession. The base ClientSession implements functions that interact with devices and other common functionality. RemoteSession implements functionality for places on top of this. RemoteSession and LocalSession implement the _get_target method, which allows the LocalSession to get targets without the need for places.
CL args have been split between remote and local. A new CL target labgrid-client-local has been made that contains only commands to run in local mode. In local mode the client will run based on the resources and drivers specified in the main target of the config file.
I also tried spinning up a local coordinator. In this case you'd have to create places, add matches, export resources and such behind the scenes, which seems like more effort and a worse design than the other approach
Currently it is only possible to use the labgrid-client tool with the distributed infrastructure - deployed exporter, coordinator, and so on.
When trying to use
labgrid-client
with only a local environment configuration where no remote place is specified and no coordinator address is supplied, the client errors out with message:IMHO it would be nice to close the feature gap of the client for local usage by either use the locally defined resources directly for some of it's commands like
io
, orconsole
, or start coordinator&exporter on-demand internally in some way.The text was updated successfully, but these errors were encountered: