Re: [leshan-dev] Refusing client registration

["J.F. Schloman" <john.schloman@xxxxxxxxx>]

I like this.  It will allow implementer flexibility using the ClientRegistry interface in case of internal issues when attempting to store, etc. :)

-JF

On Thu, Feb 26, 2015 at 8:22 AM, Simon Bernard <contact@xxxxxxxxxxxxxxx> wrote:

Currently, when an exception(checked or runtime) is raised, we don't reply with a BAD_REQUEST. It is just catched by californium which "sysout" the stack trace and never answers to the request (this will probably cause a timeout at client side).

Anyway, the changes proposed  sound good for me !

I open 2 issue at OMA to modify the spec about that :
I think when an exeception is raised we should send a 500 internal error :
https://github.com/OpenMobileAlliance/OMA-LwM2M-Public-Review/issues/36

When the registration is not accepted (registerClient returns false), we should use the 4.03 response code which seems badly described in the spec :
https://github.com/OpenMobileAlliance/OMA-LwM2M-Public-Review/issues/37

Le 26/02/2015 13:32, Julien Vermillard a écrit :

Hi,

During the shim refactoring I removed some unused checked exceptions.

One is problematic on ClientRegistry interface:

Client registerClient(Client client) throws ClientRegistrationException;

The ClientRegistrationException wasn't really explicit

/**

* An exception raised when a client registry operation has failed.

*/

So I decided to remove it because nowhere in the leshan-standalone code we was throwing it.

But apparently ClientRegistrationException was used to refuse the registration (e.g. I don't know this client).

Anyway any exception (checked or runtime) received by the registration code will just produce a BAD_REQUEST return code, the only error code we can return on the bootstrap operation...

I propose to change it to:

boolean registerClient(Client client);

returning true if the registration was accepted or not.

That's also mean we don't return anymore the previous Client instance if a Client was already registered for this endpoint. I hope this feature is not used :)

Julien

_______________________________________________
leshan-dev mailing list
leshan-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/leshan-dev

_______________________________________________
leshan-dev mailing list
leshan-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/leshan-dev