Hi Michal, On Wed, 2019-07-24 at 09:55 +0200, michal.lowas-rzechonek@xxxxxxxxxxx wrote: > Inga, > > On 07/24, Stotland, Inga wrote: > > > - ele->path = l_strdup(path); > > > > The path needs to be preserved if this is the Attach() request. > > One way to do this would be to save it here, in the temporary node > > and > > then after all the checks are done, move it to the node that is > > "owned" > > by the daemon prior to deleting the temporary node in > > get_managed_objects_cb() > (...) > > So here, prior to removing the temorary node, the element paths > > need to > > be copied into the "req->attach" version of the node. > > Same goes for node->agent and node->provisioner. > > True, thanks! I'll fix this in v3. > Also, could you please add some comment there, like "Deleting the temporary node" just to make the point that tit was temporary? I am afraid that for an outside person this will not be clear why the node is being removed. > > "Location" property is described as optional in mesh-api.txt. It's > > populated with "Default Location", if the property is not present. > > I believe, it was more out of the convenience for the app > > develpers, > > but maybe we should change the property description as mandatory, > > i.e., > > remove "optional" from the property descrition in mesh-api.txt > > I'm inclined to make this property mandatory, but I don't have a > strong > opinion about it. So just tell me how should it be and I'll adjust v3 > accordingly. > Let's try to keep it as optional then. This will require some *fuzzy* matching of the composition data, something like: add a boolean flag to generate_node_composition() that indicates whether the composition is generated for the verification or as a result of an external request, i.e. for the config server model. Based on the flag, either include location field, or not. Regards, Inga
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
