headers
Benjamin Franz | 1 Mar 1999 03:11

Re: DRAFT Mime headers 0.4

On 27 Feb 1999, Marty Fouts wrote:

> 
> Brad Templeton pounded silicon into:
> 
> [snip]
> 
> > I know of no newsreader that requires system permissions to run,
> > however I realize not everybody wants to install one in their local
> > disk space.
> 
> Nor does everyone have access to either the tools or sufficient disk space.
> 
> > However, it is worth noting that USENET has changed vastly in the
> > past few years, and users of shared hosts (what used to be called
> > timesharing) are a small minority of the net these days.
> 
> I suspect that, once you account for AOL, and for ISPs like the one I
> use, which has > 50,000 shell customers, you will discover that the
> 'small minority' is the overwhelming number of users of the net.

You are provably incorrect, at least regarding the Usenet. I ran stats
recently on the news agents of posters to soc.culture.japan. 

Here is the results of data reduction to unique posters:

N Posts: 710
   N      % News Readers (N = 16)
----------------------------------------------------------------------
   1    0.6 AOL Offline Reader
(Continue reading)

Permalink | Reply | 1 comment |
headers
John Moreno | 1 Mar 1999 04:42

Re: DRAFT Mime headers 0.4

Brad Templeton <brad <at> templetons.com> wrote:

> Trying to be backwards compatible -- which is good -- does not mean you
> cripple new functionality.
> 
> MIME took that approach with multipart/alternative, saying the plain text
> form comes first so at worst people just see some garbage at the end if
> they have the oldest readers.    They didn't just throw up their hands
> and say, "There are old mail agents out there, so we will deprecate
> non-plaintext mail."
> 
> Now I would rather that we could have gotten W3C or the MIME body to
> support an out of band encoding scheme such as the one I designed, but
> so far that has not happened.
> 
> Such schemes face a chicken and egg.   Netscape at one time said they
> were willing to implement such a scheme *if* it were an IETF or W3C
> standard.  But we need to have a working newsreader generating the
> scheme before getting that.

As I've said before the /least/ we should do is specify that formatting
information SHOULD/MUST go into the headers.  We don't have to specify a
method, just where the data should go.

--

-- 
John Moreno
Permalink | Reply | 4 comments |
headers
Martin J. Duerst | 1 Mar 1999 05:30
Picon
Favicon

Re: DRAFT Mime headers 0.4

At 22:42 99/02/28 -0500, John Moreno wrote:
> Brad Templeton <brad <at> templetons.com> wrote:

> > Now I would rather that we could have gotten W3C or the MIME body to
> > support an out of band encoding scheme such as the one I designed, but
> > so far that has not happened.
> > 
> > Such schemes face a chicken and egg.   Netscape at one time said they
> > were willing to implement such a scheme *if* it were an IETF or W3C
> > standard.  But we need to have a working newsreader generating the
> > scheme before getting that.
> 
> As I've said before the /least/ we should do is specify that formatting
> information SHOULD/MUST go into the headers.  We don't have to specify a
> method, just where the data should go.

Obviously specifying something like this without something workable
doesn't make much sense, does it.

As for the W3C, there is indeed a method that puts much of the
information out of band. The underlying principle is called
Style Sheets, and the actual technology is CSS. Please see
http://www.w3.org/Style/.

While style sheets don't remove everything but the plain text
from the document itself, they are able to remove most of the
very bothering and bloated formatting instructions that currently
are produced by many of the HTML producing tools.

What stays with the text itself is structural markup. Trying to
(Continue reading)

Permalink | Reply | 3 comments |
headers
John Moreno | 1 Mar 1999 07:03

Re: DRAFT Mime headers 0.4

Martin J. Duerst <duerst <at> w3.org> wrote:

> At 22:42 99/02/28 -0500, John Moreno wrote:
> > Brad Templeton <brad <at> templetons.com> wrote:
> 
> > > Now I would rather that we could have gotten W3C or the MIME body to
> > > support an out of band encoding scheme such as the one I designed, but
> > > so far that has not happened.
> > > 
> > > Such schemes face a chicken and egg.   Netscape at one time said they
> > > were willing to implement such a scheme *if* it were an IETF or W3C
> > > standard.  But we need to have a working newsreader generating the
> > > scheme before getting that.
> > 
> > As I've said before the /least/ we should do is specify that formatting
> > information SHOULD/MUST go into the headers.  We don't have to specify a
> > method, just where the data should go.
> 
> Obviously specifying something like this without something workable
> doesn't make much sense, does it.

Depends upon what you mean by workable -- I don't feel that we need to
specify an encoding method.  We could show one as an example, but not
bless it as the one true way  -- that's actually part of the beauty of
putting it in the headers, users don't need to worry about it, future
migration from one encoding method to another is transparent (the
newsreaders just switch from one header name to another).

We shouldn't care whether the newsreader is putting it into some type of
html or whatever it was NISUS WRITER was using on the Mac to put the
(Continue reading)

Permalink | Reply | 2 comments |
headers
Martin J. Duerst | 1 Mar 1999 07:26
Picon
Favicon

Re: DRAFT Mime headers 0.4

At 01:03 99/03/01 -0500, John Moreno wrote:
> Martin J. Duerst <duerst <at> w3.org> wrote:
> 
> > At 22:42 99/02/28 -0500, John Moreno wrote:

> > > As I've said before the /least/ we should do is specify that formatting
> > > information SHOULD/MUST go into the headers.  We don't have to specify a
> > > method, just where the data should go.
> > 
> > Obviously specifying something like this without something workable
> > doesn't make much sense, does it.
> 
> Depends upon what you mean by workable -- I don't feel that we need to
> specify an encoding method.  We could show one as an example, but not
> bless it as the one true way

Agreed in principle.

>  -- that's actually part of the beauty of
> putting it in the headers, users don't need to worry about it, future
> migration from one encoding method to another is transparent (the
> newsreaders just switch from one header name to another).

MIME headers are not exactly made to be longer than the actual
body part. Although you might be able to clean that up, too,
some older implementations may have limitations. If you separate
out formatting information, it's still better to have it in a part
of a multipart-related or whatever.

> We shouldn't care whether the newsreader is putting it into some type of
(Continue reading)

Permalink | Reply | 1 comment |
headers
Brad Templeton | 1 Mar 1999 08:27
Picon

Re: DRAFT Mime headers 0.4

On Sat, Feb 27, 1999 at 08:19:52PM -0800, Marty Fouts wrote:
> 
> I suspect that, once you account for AOL, and for ISPs like the one I
> use, which has > 50,000 shell customers, you will discover that the
> 'small minority' is the overwhelming number of users of the net.

Afraid not.  I was not speculating, I was using real numbers gathered
on the user-agent field of postings (which don't count lurkers but can
be presumed with such a huge sample space to be a pretty good guide.

We already have 50% of posters using Microsoft Outlook or Netscape.
Another 10% or so use Agent (mostly Agent 1.5) and other than the 30%
or so unidentified (which I will guess include trn -- which now supports
MIME and prettyprinted HTML, and NN).   Kjell Irgens had been collecting these
stats for years now.

So could you please tell me which facts led you to claim that the
overwhelming majority of users of the net don't have or can't upgrade
to a MIME aware newsreader?

> In addition to those users, there are also we who use Emacs, a fairly
> significant percentage of Unix users, who, as of yet, do not have a
> MIME aware newsreader, or, for that matter, reasonable MIME capability
> in our mailer.  Before you argue that such is a good reason for
> abandoning emacs, I would only offer that this is a tradeoff, and Gnus
> has many other features that more than atone for its deficiency in
> dealing with MIME, which is, after all, as of yet rare on USENET.

I am surprised nobody has done that in GNUS, but it's dwindling in share.
Perhaps because of this, but who knows?
(Continue reading)

Permalink | Reply | 38 comments |
headers
John Moreno | 1 Mar 1999 08:44

Re: DRAFT Mime headers 0.4

Martin J. Duerst <duerst <at> w3.org> wrote:

> At 01:03 99/03/01 -0500, John Moreno wrote:
> > Martin J. Duerst <duerst <at> w3.org> wrote:
> > 
> > > At 22:42 99/02/28 -0500, John Moreno wrote:
> 
> > > > As I've said before the /least/ we should do is specify that
> > > > formatting information SHOULD/MUST go into the headers.
> > > 
-snip-
> >  -- that's actually part of the beauty of putting it in the headers,
> > users don't need to worry about it, future migration from one encoding
> > method to another is transparent (the newsreaders just switch from one
> > header name to another).
> 
> MIME headers are not exactly made to be longer than the actual
> body part. Although you might be able to clean that up, too,
> some older implementations may have limitations. If you separate
> out formatting information, it's still better to have it in a part
> of a multipart-related or whatever.

These aren't MIME headers, and I vigourously disagree about multipart.
Attaching a file should be multipart, a version in french and one in
german should be multipart -- formatting shouldn't.

> > We shouldn't care whether the newsreader is putting it into some type of
> > html or whatever it was NISUS WRITER was using on the Mac to put the
> > styling information into the resource fork.
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Brad Knowles | 1 Mar 1999 09:26
Picon

Re: DRAFT Mime headers 0.4

On Sun, Feb 28, 1999, Benjamin Franz <snowhare <at> nihongo.org> wrote:

>You are provably incorrect, at least regarding the Usenet. I ran stats
>recently on the news agents of posters to soc.culture.japan.

    This is just one newsgroup, which is not a representative sample of
Usenet as a whole.  And since I believe that this sample covers a small
period of time (probably just a few days, if not only one day), I don't
believe that this is a representative sample of that newsgroup, either. 
Finally, this is just posters and not readers.  I guarantee you that
*boatloads* of more AOL users read
alt.binaries.pictures.sex.with.furry.animals than you could possibly
imagine, or that will *ever* post to newsgroups like that.

    In short, we need better stats.  Not only for the news postings
(which we can clearly read off the newsgroup itself), but also for the
news readership.  This sample needs to be taken over a long period of
time, and over a broad array of newsgroups (especially including the ones
with exceptionally high readership, such as alt.binaries.*).

--

-- 
  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|
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
                The Sky is no longer the limit
(Continue reading)

Permalink | Reply | 0 comments |
headers
Brad Knowles | 1 Mar 1999 09:19
Picon

Re: DRAFT Mime headers 0.4

On Sun, Feb 28, 1999, Seth Breidbart <sethb <at> panix.com> wrote:

>Thing to do: post binary (8-bit) file.

    Well, it's not really binary (8-bit) files, it's more like
non-plaintext, right?  What happens with other text encodings for
preserving diacriticals in Western European languages?

>Ways of doing it: binary, MIME, UUENCODE, hex, . . .

    If we're talking true binary, MIME Base64 and Uuencode are both good
ways to do it, although I'd prefer the former to the latter I think we
still have to support both.  With regards to other content, HTML probably
wouldn't have any non-ASCII characters in it, but it's not exactly what I
would call "plaintext".

    IMO, these are more complex (and overlapping) issues than I feel has
been discussed so far.

--

-- 
  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|
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
                The Sky is no longer the limit
(Continue reading)

Permalink | Reply | 1 comment |
headers
Brad Knowles | 1 Mar 1999 09:38
Picon

Re: DRAFT Mime headers 0.4

On Sun, Feb 28, 1999, Brad Templeton <brad <at> templetons.com> wrote:

>Afraid not.  I was not speculating, I was using real numbers gathered
>on the user-agent field of postings (which don't count lurkers but can
>be presumed with such a huge sample space to be a pretty good guide.

    No, this is not an accurate assumption.  Remember, I used to work at
AOL.  Although I don't have first-hand access to their statistics, I do
recall that the news readership numbers were *way* the hell skewed as
compared to the news postings, especially with regards to the pornography
and sex-related newsgroups (both the ones with pictures and the ones
without).  AOL is a large source of news postings, but they almost
certainly have the largest group of news readers on the planet.  And you
can't just blow them off, because as we have learned in the past, they
will do whatever the hell they feel like doing, and the hell with things
like standards, this WG, etc....

    And who here doesn't do boatloads of research on Dejanews, or some
other news-via-web interface?  Who's to say that all of them are fully
compliant with *anything* we're discussing?

    Well, I take that back -- Kurt Welch (newsreader.com) is pretty good
at dotting the i's and crossing the t's, and this is part of why he's got
the *only* news-via-web service that has the GNKSA.

>We already have 50% of posters using Microsoft Outlook or Netscape.
>Another 10% or so use Agent (mostly Agent 1.5) and other than the 30%
>or so unidentified (which I will guess include trn -- which now supports
>MIME and prettyprinted HTML, and NN).   Kjell Irgens had been collecting
these
(Continue reading)

Permalink | Reply | 10 comments |

Gmane