[tcpm] Re: draft-ietf-tcpm-tcp-uto-06

On 2007-10-29, at 19:33, ext Alfred HÎnes wrote:
> (1)  Terminology -- Section 3, 3.1 et al.
>
> Section 3 newly has introduced the variable 'USER_TIMEOUT':
>
>    USER_TIMEOUT
>       TCP's USER TIMEOUT parameter, as specified in [RFC0793].
>
> The subsequent text uses 'user timeout', 'USER TIMEOUT', and
> 'USER_TIMEOUT' in a way that seems to be a bit confusing; the
> distinction between these terms, and the rationale for selecting
> one of these terms remains unclear in some places.

USER TIMEOUT refers to the RFC793-defined TCP state variable.

USER_TIMEOUT names a variable in the formula in Section 3.1. that is  
used to set the local USER TIMEOUT. It's not a TCP state variable;  
USER TIMEOUT is.

The lower-case phrase is used in the body of the text when the  
concept is talked about, and not one of the specific items above. At  
least that was the intent - we may have used the lower case in some  
places where we should have used the upper case form.

> Also, the title of Section 3.1 contains 'User Timeout' whereas most
> occurrences of 'user timeout' have been changed to 'USER TIMEOUT'
> in the body of Section 3.1.

All section titles have initial capitals.

> I propose to reconsider the distinction, and to use 'USER TIMEOUT'
> only in the immediate context of RFC 793 as opposed to this draft;
> most of the occurrences of 'USER TIMEOUT' should better be replaced
> by 'USER_TIMEOUT' (the TCP state variable) or 'user timeout' (the
> abstract concept as seen from the application / user).

USER_TIMEOUT isn't the TCP state variable, USER TIMEOUT (from 793) is.

I'm OK with using the lower-case variant throughout the text, if  
people think that would eliminate confusion?

> Also, the names of the two variables, USER_TIMEOUT and LOCAL_UTO,
> are not good mnemonics for the intended purpose of these variables;
> IMHO, the naming should be reconsidered again (I already had argued
> in this matter).  In particular, 'LOCAL_UTO' now has much diminuished
> *local* significance, it it just the value advertised to the peer
> (=> my new proposal: ADV_UTO ); USER_TIMEOUT ist the important local
> state variable, and there's now only a loose coupling between both.

I like ADV_UTO, change made.

> (2)  Section 3
> In the second-to-last paragraph of Section 3, there's an unnecessary
> word replication:

fixed

> (3)  Section 3.3
> In the last paragraph, remove the duplicate full-stop in  
> "connection.."

done

Thanks!

Lars

[tcpm] Re: draft-ietf-tcpm-tcp-uto-06

Attachment: smime.p7s