headers
Brad Templeton | 1 Jan 2002 02:09
Picon

Re: Avoiding X-Headers

On Mon, Dec 31, 2001 at 05:40:03PM -0500, Henry Spencer wrote:
> Exactly.  How is the "wider net out there" supposed to know *how* to act
> on them, otherwise?
> 
> Currently, X- headers are basically used as comments.  I would rather see
> this outlawed, with explicit forms provided for specific desirable uses. 
> We don't need a whole class of headers for comments, and much of the use
> that is currently made of them is unjustifiable noise. 
> 
> I note, with some interest (and considerable surprise), that while RFC 822
> made some effort to promise that no X- header would ever be standardized,
> RFC 2822 is completely silent on the matter.

I'm confused, what do you want outlawed?  Surely not experimental new
headers?   They are the lifeblood of possible innovation in USENET.

The fact that the standard requires that unknown headers be ignored allows
such experimentation, and it's good.

The main point for me was that putting "X-" on the headers is a poor
design, and should be removed from the standard.  The standard can say
nothing, or simply suggest the person wishing to define a new header go
find the registry.  (It need not even say where the registry is if we
don't yet know.  Any decent designer will know how to use google.)

I would personally go further and throw some parameter definitions in for
new headers, to make them more useful, but now is not the time.
Permalink | Reply | 27 comments |
headers
Russ Allbery | 1 Jan 2002 02:30
Picon
Favicon
Gravatar

Re: Avoiding X-Headers

Brad Templeton <brad <at> templetons.com> writes:
> On Mon, Dec 31, 2001 at 05:40:03PM -0500, Henry Spencer wrote:

>> Exactly.  How is the "wider net out there" supposed to know *how* to
>> act on them, otherwise?

>> Currently, X- headers are basically used as comments.  I would rather
>> see this outlawed, with explicit forms provided for specific desirable
>> uses. We don't need a whole class of headers for comments, and much of
>> the use that is currently made of them is unjustifiable noise.

> I'm confused, what do you want outlawed?  Surely not experimental new
> headers?  They are the lifeblood of possible innovation in USENET.

I think Henry's arguing for the outlawing of propagation of X- headers to
Usenet as a whole.  Use X- headers for localized hacks or local
experimentation, but give the header a real name if you're going to
distribute it to all of Usenet, even as an experiment.  But I'm not sure
that I've read his argument correctly.

> The main point for me was that putting "X-" on the headers is a poor
> design, and should be removed from the standard.

It's still valuable to have a way of designating local headers.  With the
benefit of hindsight, we could have built into news software a refusal to
propagate X- headers off the local server.  Probably too late for that
now.

--

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

Permalink | Reply | 10 comments |
headers
Henry Spencer | 1 Jan 2002 02:48

Re: Avoiding X-Headers

On Mon, 31 Dec 2001, Brad Templeton wrote:
> > Currently, X- headers are basically used as comments.  I would rather see
> > this outlawed, with explicit forms provided for specific desirable uses. 
> > We don't need a whole class of headers for comments, and much of the use
> > that is currently made of them is unjustifiable noise...
> 
> I'm confused, what do you want outlawed?  Surely not experimental new
> headers?   They are the lifeblood of possible innovation in USENET.

What I want outlawed is exactly what I specified:  the user of X- headers
*AS COMMENTS*.  That is, as material deliberately intended to be read only
by humans.  There is a common practice -- which I unfortunately encouraged
in son-of-1036 -- of transforming a header, e.g. an about-to-be-superseded
NNTP-Posting-Host header, into a comment by just tacking an "X-" on the
front.  The X- prefix should be reserved for experimental new headers,
i.e. for headers intended to eventually be meaningful to software; it
should not be used as a generic comment convention. 

> The fact that the standard requires that unknown headers be ignored allows
> such experimentation, and it's good.

Agreed.  But one should distinguish intent from implementation; an unknown
header should be ignored, yes, but it should never have been inserted in
the first place if the intent was that it always remain "unknown". 

> The main point for me was that putting "X-" on the headers is a poor
> design, and should be removed from the standard.  The standard can say
> nothing, or simply suggest the person wishing to define a new header go
> find the registry...
(Continue reading)

Permalink | Reply | 12 comments |
headers
Brad Templeton | 1 Jan 2002 02:54
Picon

Re: Avoiding X-Headers

On Mon, Dec 31, 2001 at 08:48:28PM -0500, Henry Spencer wrote:
> There is an already-established convention that an X- prefix indicates an
> experimental or otherwise non-standard header.  Why, exactly, is it poor
> design to go along with this existing practice? 

As noted by myself and others, if the header becomes adopted and non
experimental, it must then be renamed.

This means that all the deployed software that was handling it can't handle
it, only new software which knows the new name.  As such, people would be
loathe to post under the new form, since the x- form would be more
widely handled!  And the transition point will never be clear.

Far better to give the header the name it will eventually have if 
adopted, and to have a version tag if new versions of the header must be
designed to be not backwards compatible for some reason.  Or, if truly
needed, you can rename it.  But you should not be forced to.

If you want it to be clear, one could imagine the defition of a standard
tag of the form version=nnn, where an nnn that is less than 1.0 indicates
experimental.
Permalink | Reply | 3 comments |
headers
Henry Spencer | 1 Jan 2002 03:02

Re: Avoiding X-Headers

On Mon, 31 Dec 2001, Brad Templeton wrote:
> > There is an already-established convention that an X- prefix indicates an
> > experimental or otherwise non-standard header.  Why, exactly, is it poor
> > design to go along with this existing practice? 
> 
> As noted by myself and others, if the header becomes adopted and non
> experimental, it must then be renamed.
> This means that all the deployed software that was handling it...

See my previous comments:  deployed software should neither generate nor
recognize X- headers.  They should be reserved strictly for *experiments*.
Experiments, by definition, are not deployed net-wide.

> Far better to give the header the name it will eventually have if 
> adopted, and to have a version tag if new versions of the header must be
> designed to be not backwards compatible for some reason.

Agreed; when it's deployment time, that's what you should do.  But the X-
convention remains useful and desirable for preliminary experimenting.
Since any X- header which appears on the net at large is, by definition,
an escaped experiment, it should always be ignored.

                                                          Henry Spencer
                                                       henry <at> spsystems.net
Permalink | Reply | 2 comments |
headers
Clive D.W. Feather | 1 Jan 2002 11:20

Re: draft-ietf-usefor-article-06 and last call

Charles Lindsey said:
>> Some Telnet clients or terminals could start to act funny if fed a
>> CSI and they are set to accept 8-bit controls.
> But are there any legitimate and defined uses for those 8 bit control
> characters of ISO 8859?

8859 doesn't defined them as control characters. It merely doesn't use them
(or the 0x00 to 0x1F space).

> Do those Telnet clients or terminals actually take
> special note of them (other than not displaying them)? If not, no problem
> arises.

I've known DEC terminals that use them; there was a general rule something
like 0xNN is equivalent to ESC '[' 0x(NN-40), so 0x84 equals ESC '[' 'D'.

--

-- 
Clive D.W. Feather  | Work:  <clive <at> demon.net>   | Tel:  +44 20 8371 1138
Internet Expert     | Home:  <clive <at> davros.org>  | Fax:  +44 20 8371 4037
Demon Internet      | WWW: http://www.davros.org | Mobile: +44 7973 377646
Thus plc            |                            |
Permalink | Reply | 0 comments |
headers
Charles Lindsey | 1 Jan 2002 16:01
Picon
Picon

Re: Avoiding X-Headers

In <Pine.BSI.3.91.1011231205935.4003B-100000 <at> spsystems.net> Henry Spencer
<henry <at> spsystems.net> writes:

>Agreed; when it's deployment time, that's what you should do.  But the X-
>convention remains useful and desirable for preliminary experimenting.
>Since any X- header which appears on the net at large is, by definition,
>an escaped experiment, it should always be ignored.

But when is "deployment time"?

Some hotheads will say "tonight, as soon as I have finished hacking this
code".

Others will say "only when a standards track RFC has been accepted by the
IESG".

And there will exist every stage you can imagine between those extremes
:-( .

--

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl <at> clw.cs.man.ac.uk      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
Permalink | Reply | 0 comments |
headers
Charles Lindsey | 1 Jan 2002 15:58
Picon
Picon

Re: Avoiding X-Headers

In <Pine.BSI.3.91.1011231204010.4003A-100000 <at> spsystems.net> Henry Spencer
<henry <at> spsystems.net> writes:

>What I want outlawed is exactly what I specified:  the user of X- headers
>*AS COMMENTS*.  That is, as material deliberately intended to be read only
>by humans.  There is a common practice -- which I unfortunately encouraged
>in son-of-1036 -- of transforming a header, e.g. an about-to-be-superseded
>NNTP-Posting-Host header, into a comment by just tacking an "X-" on the
>front.  The X- prefix should be reserved for experimental new headers,
>i.e. for headers intended to eventually be meaningful to software; it
>should not be used as a generic comment convention. 

Oh Dear!

Henry wants to outlaw X-headers as pure comments, but to allow them as
experimental headers that may one day become part of Netnews.

Brad wants exactly the opposite. Experimental headers to be non-X
headers, but apparently no problem with X-headers for local or comment usage.

I think I am slightly nearer Brad's position on this, insofar as I am
perfectly happy with the present usage of X-headers for vanity, comments,
superseded headers, etc, but would like to see some mechanism whereby
experiments that are intended to be in the field eventually can have a
non-X-header at as early a stage as possible.

The problem I have with Henry's approach is that he seems to think that
experiments using X-headers can be tried out 'live' to see that they work,
but then they should be replaced by non-X-headers at the moment they cease
to be "experiments"; i.e. at the exact point when software around the
(Continue reading)

Permalink | Reply | 1 comment |
headers
Charles Lindsey | 1 Jan 2002 15:32
Picon
Picon

Re: Avoiding X-Headers

In <Pine.BSI.3.91.1011231171933.1729B-100000 <at> spsystems.net> Henry Spencer
<henry <at> spsystems.net> writes:

>Currently, X- headers are basically used as comments.  I would rather see
>this outlawed, with explicit forms provided for specific desirable uses. 
>We don't need a whole class of headers for comments, and much of the use
>that is currently made of them is unjustifiable noise.

Hmm! I don't think I like that, but I shall comment later in this thread. 

>I note, with some interest (and considerable surprise), that while RFC 822
>made some effort to promise that no X- header would ever be standardized,
>RFC 2822 is completely silent on the matter.

Yes, but it is even worse than that. The X-header convention is not
mentioned at all! So, strictly speaking, any email with an X-header in it
is non-compliant. Even a header not defined in RFC 2822 is non-compliant
unless it is defined in some extension (e.g. in the MIME RFCs).

Well, for sure, there are (and will continue to be) lots of non-compliant
emails around (like 80% of them all). I don't think we want to go down that
route.

Anyway, perhaps Pete could comment, if he is listening.

--

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl <at> clw.cs.man.ac.uk      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 | 4 comments |
headers
greg andruk | 1 Jan 2002 21:05

Re: ietf-nntp LIST DISTR*

[Moved to a more appropriate list.  Internal server behavior really has
nothing at all to do with NNTP.]

On Tue, Jan 01, 2002 at 12:13:39PM +0000, Charles Lindsey wrote:
> Hmmm! Is it considered OK for injectors to be altering or creating
> Distribution headers without explicit consent from the poster?

Add a default distribution?  Sure.  INN still allows the poster to override
it.

> That would be contrary to the duties of an injecting agent as laid down in
> Usefor.

"The injecting agent MAY add other headers not already provided by
the poster . . ."
Permalink | Reply | 9 comments |

Gmane