[rohc] SigComp state creation and the state retention priority

["Lepine, Jean-Pierre" <jeanpierre.lepine at smisrd.com>]

Hi everybody,

 

I sent the following email almost 2 weeks ago but I didn't get any feedback on it.

I would really appreciate a follow-up on this.

Thanks.

 

-----Message d'origine-----
De : Lepine, Jean-Pierre
Envoyé : 31 mars, 2005 08:58
À : rohc at ietf.org
Objet : SigComp state creation and the state retention priority

Hi all,

 

I would just like to confirm something regarding SigComp state creation and the state retention priority.

 

Initially, I thought that trying to create a state item in a compartment could NEVER push state items with a higher priority out of the compartment. In particular, suppose a compartment uses its full amount of SMS with state items of priority 1, I thought that trying to create a state item with priority 0 would fail without any effect on the compartment's content.

 

However, after reviewing RFC 3320, I now understand that for the above example, one or more state items of priority 1 MUST be pushed out of the compartment (freed) until enough memory is available to create the state item of priority 0. More generally, a state item with a given priority, even the highest, CAN be pushed out of a compartment by the creation of state items with lower priorities. In other words, a newer state item is considered more important than an older state item, even if the older one has a higher priority.

 

Am I right ?

 

If so, I guess there are some key motivations behind this design choice but I don't really see them. I find this to be not very practical as a compressor must be careful not only with the chosen state priorities, but also with the order of state creation requests.

 

_______________________________________________
Rohc mailing list
Rohc at ietf.org
https://www1.ietf.org/mailman/listinfo/rohc