[news.software.b] Who generates the `Lines:' header field?

cbp@foster.avid.oz (Cameron Paine) (08/05/90)

I installed nn last weekend. In experimenting with it during the week, I
noted that it displays the number of lines on its subject-selection menu.
I also noted that some articles that originated on this system displayed
a '?' instead of the number of lines. Further investigation revealed that
none of the locally posted articles contained the `Lines:' header field.

Now, before I go and change something, I'd like some feedback on the related
policy. Which agent should put in the Lines: field and under what
circumstances can it be omitted? Note that I'm asking about a standard, not
about implementations.

BTW, please don't write and tell me whether this article has the field or
not. It won't have if I read it here and I intend checking the outbound
queue as soon as I've posted it to see whether the rest of the world gets
it or not.

cbp
-- 
cbp@foster.avid.oz - ACSnet
cbp%foster.avid.oz.au@uunet.uu.net - Internet
...!{hplabs,mcvax,nttlab,ukc,uunet}!munnari!foster.avid.oz.au!cbp - UUCP

henry@zoo.toronto.edu (Henry Spencer) (08/05/90)

In article <1990Aug5.064837.23881@foster.avid.oz> cbp@foster.avid.oz (Cameron Paine) writes:
>... Which agent should put in the Lines: field and under what
>circumstances can it be omitted? Note that I'm asking about a standard, not
>about implementations.

The Lines: header is generated, or not, by the news software that originated
the article (or perhaps by a gateway, if the article was composed there
based on a mail message).  It is an optional header and can be omitted
under any circumstances whatsoever.  See RFC1036.

Digressing for the moment to implementations, B News generated Lines:.
C News does not.

The size of the article, in bytes, is generally more useful and is
available to newsreaders at the cost of a query to the filesystem,
without cluttering up every article with another header.
-- 
The 486 is to a modern CPU as a Jules  | Henry Spencer at U of Toronto Zoology
Verne reprint is to a modern SF novel. |  henry@zoo.toronto.edu   utzoo!henry

wisner@hayes.fai.alaska.edu (Bill Wisner) (08/06/90)

henry@zoo.toronto.edu (Henry Spencer) writes:

>The size of the article, in bytes, is generally more useful and is
>available to newsreaders at the cost of a query to the filesystem,
>without cluttering up every article with another header.

Not over NNTP. And at any rate, it's not my news reader that cares
about the size of an article -- it's *me*. I like to have an idea of
what I'm getting into, and I generally find it easier to grok a line
count than a character count.

Bill Wisner <wisner@hayes.fai.alaska.edu> Gryphon Gang Fairbanks AK 99775
bnug, dude
yeah
.

henry@zoo.toronto.edu (Henry Spencer) (08/06/90)

In article <1990Aug5.234520.19583@hayes.fai.alaska.edu> wisner@hayes.fai.alaska.edu (Bill Wisner) writes:
>[bytes vs. lines] And at any rate, it's not my news reader that cares
>about the size of an article -- it's *me*. I like to have an idea of
>what I'm getting into, and I generally find it easier to grok a line
>count than a character count.

I find it easier still to glance at rn's percentage-of-article-yet-seen
readout, which is character-count based and thus more accurately reflects
how long it will take to splat it up on the screen.

It is admittedly a nuisance that NNTP won't supply the size of the article,
but to some extent this reflects a fundamental design bug in NNTP:  it tries
to be a reading interface AND a posting interface AND a transmission interface
all rolled into one, and consequently is not well thought out for any of
those purposes :-).  (Rev 2 may be better.)
-- 
The 486 is to a modern CPU as a Jules  | Henry Spencer at U of Toronto Zoology
Verne reprint is to a modern SF novel. |  henry@zoo.toronto.edu   utzoo!henry

jetfuel@csusac.csus.edu (Dave Jenks) (08/06/90)

In article <1990Aug5.075054.11185@zoo.toronto.edu> henry@zoo.toronto.edu (Henry Spencer) writes:
>Digressing for the moment to implementations, B News generated Lines:.
>C News does not.

I have cnews 041690 on my 3b1 (getting ready to connect to csusac),
and in inews, the block beginning at line 181, is the code to produce
the "Lines: " line.
-- 
=======================================================================
	    "Pro is to con, as progress is to Congress..."
>>-@@-> Dave Jenks		...!ucbvax!ucdavis!csusac!jetfuel
=======================================================================

wisner@hayes.fai.alaska.edu (Bill Wisner) (08/06/90)

henry@zoo.toronto.edu (Henry Spencer) writes:

>I find it easier still to glance at rn's percentage-of-article-yet-seen
>readout, which is character-count based and thus more accurately reflects
>how long it will take to splat it up on the screen.

I don't use rn. (For the record: C news with Lines: header generation
enabled and GNUS with Message-ID: header generation disabled.)

While a percentage readout is useful (and I do get one in GNUS once
I've started reading an article) I still like to have an idea of how
long an article is before I start reading it. For this, I'd rather
have a line count than a character count. It's easier for me to
mentally parse the former.

Bill Wisner <wisner@hayes.fai.alaska.edu> Gryphon Gang Fairbanks AK 99775
"Bill Wisner, how dare you send such a message!"
-- Anders G|ransson <ag@sics.se>

brian@ucsd.Edu (Brian Kantor) (08/06/90)

In article <1990Aug6.024749.28260@zoo.toronto.edu> henry@zoo.toronto.edu (Henry Spencer) writes:
>It is admittedly a nuisance that NNTP won't supply the size of the article,
>but to some extent this reflects a fundamental design bug in NNTP:  it tries
>to be a reading interface AND a posting interface AND a transmission interface
>all rolled into one, and consequently is not well thought out for any of
>those purposes :-).  (Rev 2 may be better.)

It won't be.

henry@zoo.toronto.edu (Henry Spencer) (08/07/90)

In article <1990Aug6.061150.3939@csusac.csus.edu> jetfuel@csusac.UUCP (Dave Jenks) writes:
>>Digressing for the moment to implementations, B News generated Lines:.
>>C News does not.
>
>I have cnews 041690 on my 3b1 (getting ready to connect to csusac),
>and in inews, the block beginning at line 181, is the code to produce
>the "Lines: " line.

It's there but commented out.  People who really want the silly header
can uncomment it.  We're not going to make it the default, though.
-- 
The 486 is to a modern CPU as a Jules  | Henry Spencer at U of Toronto Zoology
Verne reprint is to a modern SF novel. |  henry@zoo.toronto.edu   utzoo!henry

cbp@foster.avid.oz (Cameron Paine) (08/10/90)

In <1990Aug5.075054.11185@zoo.toronto.edu>, I asked about policy relating
to the Lines header field. Fortunately, my question didn't spark a flame
war and the responses (both postings and e-mail) fell into three
categories: policy, personal preference and solutions.

The policy seems to be clear. Lines is an optional field and if it is to
be included, it should be supplied by the posting agent. Henry Spencer
pointed out that B News does this and C News doesn't. He contended that
the field was unnecessarily wasteful. Interestingly, his article had a
Lines field.

On reflection, I believe that Henry's rationale makes sense. Assuming
articles have on average N lines (where 10 < N < 100), the field would
require strlen("Lines: 99\n") (ie 10) characters. Multiply this by the
number of articles received and stored each day and we're looking at a
small but significant volume of data.

As an aside, C News article IDs are more than 10 characters longer than
their B News counterparts (on most installations). Thus, the size of a
C News header is larger than a similar B News header. :-)

Responses in the personal preference category related to the user
interface of the reading agent. Henry Spencer suggested that character
counts were a better article size metric because they could be cheaply
determined by the reading agent. An alternative view was advanced by
Bill Wisner (et al) who suggested that line counts were easier to
comprehend than character counts.

Several people told me how to turn on the Lines computation in C News.
Thanks, but I had already Read The Fine Sources.

From all this I conclude that article size computation (in whatever
form) can be more efficiently determined by the reading agent. I will
modify the nn sources to implement this policy and will post the patches
to Kim Storm for possible inclusion in a future release.

cbp
-- 
cbp@foster.avid.oz - ACSnet
cbp%foster.avid.oz.au@uunet.uu.net - Internet
...!{hplabs,mcvax,nttlab,ukc,uunet}!munnari!foster.avid.oz.au!cbp - UUCP

henry@zoo.toronto.edu (Henry Spencer) (08/11/90)

In article <1990Aug10.164353.6188@foster.avid.oz> cbp@foster.avid.oz (Cameron Paine) writes:
>... Interestingly, his article had a Lines field.

It wasn't supplied here!  Some (all?) B Newses rewrite headers for you,
so the article presumably passed through such a site on its way to you.
Geoff and I consider unnecessary header rewriting to be a major sin, and
refuse to do it.

>As an aside, C News article IDs are more than 10 characters longer than
>their B News counterparts (on most installations). Thus, the size of a
>C News header is larger than a similar B News header. :-)

This will be fixed. :-)
-- 
It is not possible to both understand  | Henry Spencer at U of Toronto Zoology
and appreciate Intel CPUs. -D.Wolfskill|  henry@zoo.toronto.edu   utzoo!henry

chuck@dworkin.wustl.edu (Chuck Cranor) (08/14/90)

In article <1990Aug11.031036.11586@zoo.toronto.edu> henry@zoo.toronto.edu (Henry Spencer) writes:
>In article <1990Aug10.164353.6188@foster.avid.oz> cbp@foster.avid.oz (Cameron Paine) writes:
>>... Interestingly, his article had a Lines field.

>It wasn't supplied here!  Some (all?) B Newses rewrite headers for you,
>so the article presumably passed through such a site on its way to you.
>Geoff and I consider unnecessary header rewriting to be a major sin, and
>refuse to do it.

Oh, I consider it a feature.  I like all my news presented in the same
format and header order.  I guess you could argue that such tasks should
be done by a news reader.  However my news reader (rn/trn) doesn't
seem to do that.  Until recently I've been stuck on a C news system
with my headers in the "wrong" order, no "Lines:", and I found it 
really annoying.

I guess you see it differently, but in my mind by making C news behave
this way you've broken a standard feature which I was used to in 
order to bully me into upgrading my news reading software. :-) 

cheers,
Chuck

--
Chuck Cranor
E-Mail: chuck@maria.wustl.edu / cranor@udel.edu
here: 8069 Valcour Ave, Apt 202, St. Louis MO 63123
home: 104 Ashley Court, Greenville DE, 19807