RFC: Partly fix handling of mountpoints in the 'up' command #519
Conversation
The cli_auto_up command right now does not properly handle mountpoints. Fix the case when going up one level inside of a mounted module. Note that this does not yet fix the case where the current level is the top-level of the mounted module. In that case the up command should switch to the presence container that holds the mounted module. Instead it switches all the way up to the topmost level, so it has the same effect as the top command.
|
I know,there is a lot of newautocli code in the controller, that needs to be retrofitted into more generic clixon code. But I never got edit modes to work, so the submission is appreciated. |
|
Thanks for merging this!
Do you remember any specific problems in this area? For me edit mode basically appears to work in combination with mountpoints. |
|
Operation across mountpoints, such as "up" |
Not really but there is an issue in the controller project: |
|
I don't remember having seen error messages similar to the one in the mentioned clixon-controller issue. However, I did get some error-messages before due to inconsistent cli-edit-* states caused by earlier calls to up and top commands. But these should be addressed now. |
|
We will revisit that issue and see if it works now. |
The
upcommand right now does not properly handle mountpoints. Partly fix that.Note that this does not yet fix the case where the current level is the top-level of the mounted module. In that case the
upcommand should switch to the presence container that holds the mounted module. Instead it switches all the way up to the topmost level, so it has the same effect as thetopcommand.Fixing that would be more involved:
co_up()on the top-levelcg_objof a mounted module returnsNULL. That case would need to be handled by finding thecg_objmatching the mount-point container, i.e. the path stored in"cli-edit-mtpoint", and setting that as the newcligen_ph_workpoint. But I don't now how to get that cg_obj.