Re: [netmod] tree diagrams - flags

[Martin Bjorklund <[e-mail]>]

Juergen Schoenwaelder <[e-mail]> wrote:
> Hi,
> 
> if we want to standardize tree diagrams, we may want to take a more
> critical look at them, in particular the flags (that were created
> ad-hoc and in resemblance to MIB tree diagrams). pyang --tree-help
> says:
> 
>   <flags> is one of:
>     rw  for configuration data
>     ro  for non-configuration data
>     -x  for rpcs and actions
>     -n  for notifications
> 
> This is (a) incomlete and (b) somewhat confusing since ct does not
> equate to readwrite. I am attaching a sample yang file and here is the
> output pyang 1.7.1 produces:
> 
> module: tree-sample
>     +--rw config-true-container
>     |  +--rw param?   string
>     +--ro config-false-container
>     |  +--ro value?   string
>     +--rw inline-action
>     |  +---x action
>     |     +---- oops?     string
>     |     +---w input
>     |     |  +---w in?   string
>     |     +--ro output
>     |        +--ro out?   string
>     +--rw inline-notification
>        +---n notification
>           +---- duration?   string
> 
>   rpcs:
>     +---x rpc
>        +---w input
>        |  +---w in?   string
>        +--ro output
>        |  +--ro out?   string
>        +--ro oops?     string
> 
>   notifications:
>     +---n notification
>        +--ro boom?   string
> 
> I think a better usage of two letter flags would have been this (since
> it more naturally aligns with what the YANG definition says):
> 
>   <flags> is one of:
>     ct  for configuration data
>     cf  for non-configuration data
>     x-  for rpcs and actions
>     xi  for rpc or action input
>     xo  for rpc or action output
>     n-  for notifications
>     nt  for notification tree (this is I think the term 7950 uses)

I'm fine with this, but perhaps use "no" for notification data - "t"
means "true" in "ct".

Also, in a grouping like this:

 grouping my-grouping {
    leaf param { type string; }
  }

pyang prints this as:

  my-grouping
      +---- param?   string

i.e., w/o any flags.


> (And I think the oops leafs should have triggered an error.)

They did.  To stderr.


/martin