ietf-nntp CONSENSUS POLL -- PLEASE READ AND RESPOND NO LATER THAN 1400 CST 17 March 2002 (that +6:00 GMT for those that don't know what CST is)

[Stan O Barber]

Russ Allbery wrote:

> Stan O Barber <sob at academ.com> writes:
> 
> 
>>Overview
>>
> 
>>In the response to the OVER command, should bytes and lines be calculated
>>in real-time using the entire article (body and headers)?
>>
> 
> I'm not sure I understand this question.  I don't believe that anyone is
> arguing that the byte and line counts of the article should be calculated
> in response to the OVER command, only that the values stored in the
> overview database of the server be calculated from the article (probably
> at the time that it was received from a peer) rather than based on the
> Lines header of the message.



"real-time" means in a manner that involves direct calculation of the 
item (lines or bytes). This could be done at the time the OVER command 
is invoked or by looking a previously calculated value up in a database 
(assuming that the news article spool is inviolatile with respect to an 
article size). It is explicitly NOT found by looking at header lines. I 
realize that "real-time" was probably not the best choice of phrase 
here, but I do want the value returned (if people AGREE) to reflect the 
actual size of the article (in lines or bytes, depending) that a client 
could expect the article to be should the client retrieve it via an 
ARTICLE command.


> 
> 
>>Should blank mandatory headers return a value of ""?
>>
> 
> I'm also not sure I understand this question.  Do you mean a literal value
> of two ASCII double quote characters, or an empty value (delimited by
> tabs or the end of the overview string)?  Is this question meaning to ask
> about the representation of trailing empty fields in the return of OVER or
> something else?


I mean an empty value. It appeared from reviewing the archives that 
there was some question about when to return an empty field for 
manditory headers and when not to. Someone said that USEFOR was trying 
to insure that no header was empty, so the question may be moot, but I 
wanted to find out if there was any particular consensus on what to do 
about manditory headers that were missing or blank in an OVER response.

By the way, I left consensus on HDR out because I could not determine a 
clear question from the discussion. I could have missed something here, 
so please feel free to let me know.