Behavior of rootpath update (rt="oma.lwm2m")

[sbernard31]

The specification defines a way to put the LWM2M tree under a custom root path by using rt="oma.lwm2m" attribute at registration time. (transport§6.4.1 Alternate Path)

This looks like :

</customRootPath>;rt="oma.lwm2m", </customRootPath/1/0>,</customRootPath/3/0>

I understand that

</>;rt="oma.lwm2m", </1/0>,</3/0>

is equivalent to :

</1/0>,</3/0>

What is no clear to me is about Update Request.
If you are using a custom root path is it mandatory to put it in each update request ?
If not, what is the expected behavior.

example 1) :

Register : </customRootPath>;rt="oma.lwm2m", </customRootPath/1/0>,</customRootPath/3/0>
Update  :  </customRootPath>;rt="oma.lwm2m", </customRootPath/1/0>,</customRootPath/3/0>,</customRootPath/4/0>

example 2):

Register : </customRootPath>;rt="oma.lwm2m", </customRootPath/1/0>,</customRootPath/3/0>
Update  :  </customRootPath/1/0>,</customRootPath/3/0>,</customRootPath/4/0>

example 3):

Register : </customRootPath>;rt="oma.lwm2m", </customRootPath/1/0>,</customRootPath/3/0>
Update  :  </1/0>,</3/0>,</4/0>

For 1), it seems to be the obvious way.
For 2) I guess this is an invalid update request ?
For 3) I guess this changes the rootpath to / ?

[mlasch]

I came across this issue, because this might not be handled correctly in the Wakaama Server. Wakaama currently requires </>;rt="oma.lwm2m", and rejects registrations without having rt= set to oma.lwm2m.

I understand that

</>;rt="oma.lwm2m", </1/0>,</3/0>

is equivalent to :

</1/0>,</3/0>

If thats the case (which is not explicitly defined in the specification) we should modify Wakaama to also accept registrations without rt= set for </>. There are already workarounds, for example in the Zephyr LwM2M client, to add rt= for the default root path in order to work properly with Wakaama.

@sbernard31 Regarding your original question, I agree with your conclusions in 1), 2) and 3).

[sbernard31]

@mlasch Thx for sharing 🙏 .

It would be great to get confirmation from OMA.

[mkgillmore]

@sbernard31 is this still an issue for your implementation that needs clarification from OMA?

[sbernard31]

@mkgillmore, yes please clarification would be appreciated 🙏.

[eliasweingaertner]

My opinion:

1. valid
2. invalid, as it misses the custom rootpath. We can't expect devices to guess it from Object links
3. valid

[mkgillmore]

@sbernard31 @eliasweingaertner Comment is consensus of the group. Can I close the issue?

[sbernard31]

@mkgillmore

It sounds good to me but to be sure and before to close.
Could you clarify the meaning of example 3. 🙏 ? Do I have the "right" interpretation : "this UPDATE means that device has now a / as rootpath?