headers
Henry Spencer | 1 Feb 1999 03:06

Re: The "Lines:" header?

>     Here's something out of news.software.nntp today -- what do we do
> about the "Lines:" header?

It's a stupid idea and always has been.  It's even more stupid and useless
in the presence of MIME.  Junk it. 

>     Is it an initially generated header which must remain immutable, and
> therefore if one gets any messages with a "Lines:" header that doesn't
> match the actual number of lines in the message body the message SHOULD
> then be refused or junked?

Very bad idea.  Slightly-incorrect Lines headers have been a feature of
news almost from the header's first appearance.

>     Is it a locally modifiable header which is only intended to give the
> newsreader more useful information about the size of an article, and
> therefore if any messages come in with munged "Lines:" headers they
> SHOULD then be corrected?

While correcting it seems harmless, it's of such limited use that I doubt
it's worth breaking the rule of "never mess with headers unnecessarily". 
The overview database supplies the same information to the newsreaders,
and (server willing) does it for all articles, not just those which happen
to have the header.

>     If not one of the two above choices, then what good is it, and why
> not just deprecate it and be done with it?

It's useless.  It should be deprecated, or even completely excluded.
(Continue reading)

Permalink | Reply | 8 comments |
headers
Russ Allbery | 1 Feb 1999 05:29
Picon
Favicon
Gravatar

Re: The "Lines:" header?

Henry Spencer <henry <at> spsystems.net> writes:

>> Here's something out of news.software.nntp today -- what do we do about
>> the "Lines:" header?

> It's a stupid idea and always has been.  It's even more stupid and
> useless in the presence of MIME.  Junk it.

Yeah, I have to say this makes the most sense.  The server can trivially
generate that information itself when it needs it (for overview and for
some spam filtering).  It has a lot of the same problems as Content-Length
in mail.

--

-- 
Russ Allbery (rra <at> stanford.edu)         <URL:http://www.eyrie.org/~eagle/>
Permalink | Reply | 1 comment |
headers
Brad Knowles | 1 Feb 1999 09:39
Picon

Re(2): The "Lines:" header?

On Sun, Jan 31, 1999, Henry Spencer <henry <at> spsystems.net> wrote:

>While correcting it seems harmless, it's of such limited use that I doubt
>it's worth breaking the rule of "never mess with headers unnecessarily".

    It's of use to newsreaders, which can present that information to the
user, or perhaps even make their own determinations as to whether or not
to display the article or how to mark it.  Since this seems to be the
only place where it has any really strong value (a horribly wrong
"Lines:" header is a weak indicator that the message might have been
munged in transit, and perhaps could/should be refused), it would seem to
me that one should be able to safely modify it to be correct.

>The overview database supplies the same information to the newsreaders,
>and (server willing) does it for all articles, not just those which happen
>to have the header.

    So when a newsreader issues the XOVER command, it's getting the
"real" data out of the overview database, and not just a dump of the
headers?  That doesn't match with my experience, but maybe you can
convince me that I'm wrong.

--

-- 
  These are my opinions -- not to be taken as official Skynet policy
 ____________________________________________________________________
|o| Brad Knowles, <blk <at> skynet.be>            Belgacom Skynet NV/SA |o|
|o| Systems Architect, News/mail/FTP Admin   Rue Col. Bourg, 124   |o|
|o| Phone/Fax: +32-2-706.11.11/12.49         B-1140 Brussels       |o|
|o| http://www.skynet.be                     Belgium               |o|
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
(Continue reading)

Permalink | Reply | 5 comments |
headers
Russ Allbery | 1 Feb 1999 10:20
Picon
Favicon
Gravatar

Re: The "Lines:" header?

Brad Knowles <blk <at> skynet.be> writes:
> On Sun, Jan 31, 1999, Henry Spencer <henry <at> spsystems.net> wrote:

>> While correcting it seems harmless, it's of such limited use that I
>> doubt it's worth breaking the rule of "never mess with headers
>> unnecessarily".

> It's of use to newsreaders, which can present that information to the
> user, or perhaps even make their own determinations as to whether or not
> to display the article or how to mark it.

Newsreaders should be doing this on the basis of overview information,
which can be generated without adding the header to articles.

> So when a newsreader issues the XOVER command, it's getting the "real"
> data out of the overview database, and not just a dump of the headers?

Correct on all the server implementations I've personally worked with.  In
INN, for example, the overview database is generated by the overchan
process, which receives a special feed of information from innd.  innd can
easily be made to include an accurate line count in that information
without requiring it be in an article header.

For evidence to support this, note that the overview format includes a
bytes value giving the size of the article in bytes; this isn't from any
header actually in the article.

--

-- 
Russ Allbery (rra <at> stanford.edu)         <URL:http://www.eyrie.org/~eagle/>
(Continue reading)

Permalink | Reply | 3 comments |
headers
Charles Lindsey | 1 Feb 1999 12:34
Picon
Picon

Re: Serious problem with the "Replaces" header

In <v04020a0cb2d731d8e572 <at> [130.237.150.138]> Jacob Palme <jpalme <at> dsv.su.se> writes:

>According to my suggestion, a parameter should be specified
>for "Supersedes" with values like "minor", "major",
>"repeat" (=faq). Depending on the architecture, the
>interpretation of this parameter can be done either in the
>server or the client. If messages are stored in a server
>jointly for several users, like in netnews and in IMAP used
>for some bulletin boards, then these parameters are
>interpreted once and in the same way for all users. If
>messages are downloaded to the client, as is common in mail
>and occurs also in news, then it is natural to interpret
>these parameters under control of the individual recipient.

I am not clear what you are proposing. What you say is what is agreed
already (except that the three words proposed are not the same as yours). 
The intended implementation is that the "server" cancels the old version
(subject to site policies on cancels in general) and the client decides
whether to view the new version or not (based on information in the Xref,
or any other way it figures out).

Note that any so-called "client" that maintains a substantial cache is, to
that extent, already a "server". So what? We shouldn't care so long as it
is not feeding anyone else.

--

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Email:     chl <at> clw.cs.man.ac.uk  Web:   http://www.cs.man.ac.uk/~chl
Voice/Fax: +44 161 437 4506      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9     Fingerprint: 73 6D C2 51 93 A0 01 E7  65 E8 64 7E 14 A4 AB A5
(Continue reading)

Permalink | Reply | 0 comments |
headers
Charles Lindsey | 1 Feb 1999 12:46
Picon
Picon

Re: The "Lines:" header?

In <19990129194128.025005 <at> relay.skynet.be> Brad Knowles <blk <at> skynet.be> writes:

>Folks,

>    Here's something out of news.software.nntp today -- what do we do
>about the "Lines:" header?

>    Is it an initially generated header which must remain immutable, and
>therefore if one gets any messages with a "Lines:" header that doesn't
>match the actual number of lines in the message body the message SHOULD
>then be refused or junked?

No. If you want a check that the text has not been munged, then use
Content-MD5.

>    Is it a locally modifiable header which is only intended to give the
>newsreader more useful information about the size of an article, and
>therefore if any messages come in with munged "Lines:" headers they
>SHOULD then be corrected?

More or less. Agents regularly do correct it (even if they shouldn't).
Some newsreaders (e.g. nn) look at it.

>    If not one of the two above choices, then what good is it, and why
>not just deprecate it and be done with it?

We could do. But we cannot abolish it because it exists, and is widely
generated and even used. "Mostly harmless" is the phrase that come to
mind.
(Continue reading)

Permalink | Reply | 10 comments |
headers
Brad Knowles | 1 Feb 1999 14:30
Picon

Re(2): The "Lines:" header?

On Mon, Feb 1, 1999, Charles Lindsey <chl <at> clw.cs.man.ac.uk> wrote:

>>    What I saw of section 6.11 of
>><http://www.ietf.org/internet-drafts/draft-ietf-usefor-article-01.txt>
>>didn't seem to be particularly useful to me.
>
>It warns against relying on it, and generally reinforces my "more or less"
>comment above. It ain't particularly broke ...

    I'm sorry, that's just wishy-washy and encourages people to interpret
the header differently (some people insist that it is generated at
injection and afterwards is inviolable, others insist that it can and
should be corrected at the news server).  This kind of behaviour
encourages multiple differing *incompatible* implementations, and
generally just confuses the issue.

    Being wishy-washy on issues like this is the absolute worst possible
thing we could do.  We need some clear language that says what this
header may and may not be used for, even if that language simply says
"Usage varies.  No news client or news server may depend on it
functioning in any particular manner; considered deprecated."

--

-- 
  These are my opinions -- not to be taken as official Skynet policy
 ____________________________________________________________________
|o| Brad Knowles, <blk <at> skynet.be>            Belgacom Skynet NV/SA |o|
|o| Systems Architect, News/mail/FTP Admin   Rue Col. Bourg, 124   |o|
|o| Phone/Fax: +32-2-706.11.11/12.49         B-1140 Brussels       |o|
|o| http://www.skynet.be                     Belgium               |o|
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
(Continue reading)

Permalink | Reply | 0 comments |
headers
Henry Spencer | 1 Feb 1999 15:24

Re: The "Lines:" header?

> >...why not just deprecate it and be done with it?
> 
> We could do. But we cannot abolish it because it exists, and is widely
> generated and even used...

We don't have to acknowledge or bless every stupid kludge which happens to
exist and be used.  While we probably do have to say something about it,
I think it's entirely reasonable to say "don't bother".

                                                          Henry Spencer
                                                       henry <at> spsystems.net
                                                     (henry <at> zoo.toronto.edu)
Permalink | Reply | 9 comments |
headers
Henry Spencer | 1 Feb 1999 15:32

Re: Re(2): The "Lines:" header?

> >The overview database supplies the same information to the newsreaders,
> >and (server willing) does it for all articles, not just those which happen
> >to have the header.
> 
>     So when a newsreader issues the XOVER command, it's getting the
> "real" data out of the overview database, and not just a dump of the
> headers?  That doesn't match with my experience...

As others have noted, the overview database contains things like the byte
count, which are not in the headers at all.  The original definition of
the format specifies that field as (optionally) containing the "line
count", not the contents of the Lines header.  It is reasonable, although 
not explicitly mandatory, for the server to count the lines itself.

                                                          Henry Spencer
                                                       henry <at> spsystems.net
                                                     (henry <at> zoo.toronto.edu)
Permalink | Reply | 0 comments |
headers
Dan Ritter | 1 Feb 1999 16:36

New Draft will be available today


I am reading over the latest version of usenet-article-format today
with an eye towards releasing it tonight.

At that time, I will send a notice telling where to get it. I would
very much appreciate the following services from list members:

- if you are the author of some portion of the draft, please
  make sure I have your latest wording, or, if I have edited it,
  that we have your meaning intact.

- that internal section references are correct. (I think they are,
  but I'm sure something will have bitten us by now.)

- if something appears to be missing, notify me immediately. If I
  have it, I will insert; if I do not, I will ask for it; if no one
  has written it, I will be compiling a list of those issues for the
  group to discuss.

Thanks,

-dsr-
Permalink | Reply | 0 comments |

Gmane