RE: [dccp] SHOULD vs. MUST in the Port Numbers section

["Phelan, Tom" <tphelan at sonusnet.com>]

Hi Eddie,

There's been a lot of discussion recently that if an RFC says SHOULD, it MUST also say the circumstances that would allow violating the SHOULD.  I forget who the police for this is supposed to be, but I have seen some recent submissions bounced back because of this.

So, if we keep the SHOULD/SHOULD NOT language, we need to add something about when violations are allowed.  Something global like "exceptions to these requirements are at IANA's discretion" might work.

That being said, I think that I'd like to see the service code requirement as a MUST.  There's a confusing overlap in the port and service code functions, and if we allow developers an out to just stick with what's familiar, we won't get the non-overlapping benefits.

Tom P.

> -----Original Message-----
> From: dccp-bounces at ietf.org [mailto:dccp-bounces at ietf.org]On Behalf Of
> Eddie Kohler
> Sent: Tuesday, November 29, 2005 2:13 AM
> To: dccp group
> Subject: [dccp] SHOULD vs. MUST in the Port Numbers section
> 
> 
> Hi all,
> 
> There are a number of SHOULD/SHOULD NOTs in the Port Numbers section  
> that could be MUST/MUST NOTs; specifically, these four:
> 
>      o  A short English phrase describing the port's purpose.  This
>         SHOULD include one or more space-separated textual 
> Service Code
>         descriptors naming the port's corresponding Service Codes...
> 
>      o  A port name SHOULD NOT be registered for more than 
> one DCCP port
>         number.
> 
>      o  Any such registration [of a port name registered for UDP]
>         SHOULD use the same port number as the existing UDP  
> registration.
> 
>      o  However, if such a registration [of a TCP/SCTP port 
> name] makes
>         sense ... the DCCP
>         registration SHOULD use the same port number as the existing
>         registration.
> 
> I left them as SHOULDs because I did not want to imply that IANA  
> should check them, and because in some cases UDP and TCP themselves  
> do not follow these rules.
> 
> Anyone have any feelings about changing these four 
> SHOULD/SHOULD NOTs  
> to MUST/MUST NOTs?  I'd default to leaving them as SHOULD/SHOULD NOTs.
> Eddie
> 
> 
>