headers
Sylvan Butler | 1 May 1997 02:45
Picon

Re: Invalid "From:" address

Peter Heirich wrote:
>Sylvan Butler wrote:
>>Whether you filter upon receipt or post with an invalid address the 
>>only difference is where in the mail transmission process the message 
>>is disgarded.  A filter places the cost upon the recipient whereas the 
>>invalid address places the burden on the sender.  (IMO the sender is 
>>the appropriate place.)
>>
>
>I don't agree, because this means a lot of mail to the postmaster. Because
>because the user dosen't exist at the site, it goes to
>postmaster <at> heirich.in-berlin.de too.

I am not sure what you mean.  In either case (when sending from user <at> X 
to invalid <at> Y then postmaster <at> Y gets the error OR postmaster <at> X gets 
the error), I find it hard to believe that any site which has more 
than a couple of users would not have changed that configuration by 
now.  None of the machines I use sends those messages to the 
postmaster.  Neither of the machines I administer sends them either, 
and I did not have to change the defaults.

Now if you mean that sending from invalidx <at> X to invalidy <at> Y and 
some postmaster gets the message then YES.  If it can't bounce back 
to a valid mailbox then one of the systems is going to quit bouncing 
it and give it to a postmaster.  Too many hops, etc.  But that 
requires someone sending mail to an invalid address from an invalid 
address.

>Filter upon the recipient is the right place.
(Continue reading)

Permalink | Reply | 5 comments |
headers
Jacob Palme | 6 May 1997 21:18
Picon
Picon

Where Usenet News can be better

The traditional way of making standards in Internet is to only make
standards on things already implemented, or of very well-known and
well-defined new functionality. There are exceptions, however, the HTTP 1.1
proposals contains more than just this.

In the case of Usenet News, there is a large set of well-known extensions
which many news systems support, so there is an obvious need for
standardising these.

Since this is the beginning of a new working group, it might be reason to
think a little wider and further than this, even though I am well aware that
such wide and far thinking is against Internet traditions.

I have been involved with computer conferencing for more than twenty years,
started out with Forum-Planet and EIES in the 70-s, and have since then
been working in a large number of projects involving computer conferencing,
like the KOM system, the EU-funded GILT and AMIGO projects, ISO work on
group communication and the EU-funded Web4Groups project. Looking back on
this experience, and comparing with Usenet News, here is a summary of what
I think are the most important facilities lacking in Usenet News:

Only always full replication
----------------------------

Usenet News is based on full replication of all documents to each server,
and that a user will normally get only information which is already
available in his local server. This system has many advantages, but it has
one large disadvantage: The volume of information is so large, that many
news servers only keep documents for a week. This is somewhat alleviated by
a few large archiving servers which keep documents for a long time.
(Continue reading)

Permalink | Reply | 33 comments |
headers
Brad Templeton | 6 May 1997 20:42
Picon
Favicon

Re: Where Usenet News can be better

On May 06, 1997 at 08:18:58PM +0100, Jacob Palme wrote:
> Large documents are split into a short textual abstract and large
> attachments. The textual abstracts are kept longer in the data base, the
> attachments are purged faster. If a user accesses a document, whose

Well, I don't know if you can get people to write abstracts of their
articles, though it would be nice.  But a number of other solutions have
been proposed to this that don't require abstract writing, but simply
delay the first fetch by the first user at a site for articles in unpopular
groups that have not been widely read.

Ie. a cache scheme where headers are sent but not bodies, but on popular
groups bodies are sent, and then some mechanism notices which articles are
getting fetched for the cache a lot and pre-feeds those as well.

> This is much more difficult to solve, because to allow closed groups, there
> is a need for a registry of members of groups, and one main principle of
> Usenet News is that no such member registries exist. To add such registries
> is a large and complex change to Usenet News, and introduces complication
> which is probably not wanted. On the other hand, there is an obvious need
> for this.

There is, and you can even do this without independent servers, but the
truth is that distributing this is probably not a good idea.  You don't
want to distribute administration of closed groups, it's a nightmare.

Chances are it's best to just say closed groups are done with controlled
hierarchies (when it's machines you want to limit, not people) and if it's
people you want to limit, done with mailing lists or web sites or central
news servers.
(Continue reading)

Permalink | Reply | 13 comments |
headers
Jacob Palme | 7 May 1997 18:34
Picon
Picon

Re: Where Usenet News can be better

At 11.42 -0700 97-05-06, Brad Templeton wrote:
> On May 06, 1997 at 08:18:58PM +0100, Jacob Palme wrote:
> > Large documents are split into a short textual abstract and large
> > attachments. The textual abstracts are kept longer in the data base, the
> > attachments are purged faster. If a user accesses a document, whose
>
> Well, I don't know if you can get people to write abstracts of their
> articles, though it would be nice.  But a number of other solutions have
> been proposed to this that don't require abstract writing, but simply
> delay the first fetch by the first user at a site for articles in unpopular
> groups that have not been widely read.

If a large document does not have an abstract, the first 50 lines could be
handled as an abstract, so that these are saved a longer time than the full
message, and the full message is retrieved from either the original server
or an archiving server.

Note that MIME already has formatting standards for all this, so Usenet
News should preferably not invent new formats for this, but specify which
subset of MIME to use.

> > (1) Define a standard for how the same document, when sent to both e-mail
> > and news, should look in each environment. The most important item in this
> > standard is that the "Newsgroups" header should be renamed "Posted-To" in
> > the e-mail version of a document (since at present, "Newsgroups" in e-mail
> > is used in two conflicting ways, and there is no change of getting each
> > side to accept the other sides view on how this header should be used).
> > John Stanley has written a draft such document, I intend to work on it and
> > send it in as an Internet Draft soon.
>
(Continue reading)

Permalink | Reply | 18 comments |
headers
Clive D.W. Feather | 8 May 1997 07:40

Re: Where Usenet News can be better

Jacob Palme said:
> If a large document does not have an abstract, the first 50 lines could be
> handled as an abstract,

Result: an archive full of quoted text from people who can't be bothered to
edit when replying.

--

-- 
Clive D.W. Feather     | Work:  <clive <at> demon.net>  | Tel: +44 181 371 1138
Director of            | Home:  <clive <at> davros.org> | Fax: +44 181 371 1037
  Software Development | Abuse: <clive <at> bofh.org>
Demon Internet Ltd.    |
Permalink | Reply | 0 comments |
headers
Jacob Palme | 9 May 1997 11:39
Picon
Picon

Re: Where Usenet News can be better

At 06:40 +0100 97-05-08, Clive D.W. Feather wrote:
> Jacob Palme said:
> > If a large document does not have an abstract, the first 50 lines could be
> > handled as an abstract,
>
> Result: an archive full of quoted text from people who can't be bothered to
> edit when replying.

My e-mail software (Eudora) can be set up so that if I get a message larger
than X kilobytes (X settable by me) I only get the first 40 lines of the
message downloaded automatically. I can then when reading these 40 lines
decide whether I want the whole message downloaded or not. I have never had
any problems deciding whether I want the whole message or not based on the
first 40 lines provided by Eudora.

------------------------------------------------------------------------
Jacob Palme <jpalme <at> dsv.su.se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/~jpalme
Permalink | Reply | 16 comments |
headers
Clive D.W. Feather | 9 May 1997 11:31

Re: Where Usenet News can be better

Jacob Palme said:
>>> If a large document does not have an abstract, the first 50 lines could be
>>> handled as an abstract,
>> Result: an archive full of quoted text from people who can't be bothered to
>> edit when replying.
> My e-mail software (Eudora) can be set up so that if I get a message larger
> than X kilobytes (X settable by me) I only get the first 40 lines of the
> message downloaded automatically. I can then when reading these 40 lines
> decide whether I want the whole message downloaded or not. I have never had
> any problems deciding whether I want the whole message or not based on the
> first 40 lines provided by Eudora.

So ?

The original suggestion was to keep abstracts around much longer and in
many more places than full articles. Looking at Usenet, I suggest that
using the first 40 lines as a default abstract would be useless.

--

-- 
Clive D.W. Feather     | Work:  <clive <at> demon.net>  | Tel: +44 181 371 1138
Director of            | Home:  <clive <at> davros.org> | Fax: +44 181 371 1037
  Software Development | Abuse: <clive <at> bofh.org>
Demon Internet Ltd.    |
Permalink | Reply | 0 comments |
headers
Jacob Palme | 9 May 1997 16:27
Picon
Picon

Re: Where Usenet News can be better

At 10:31 +0100 97-05-09, Clive D.W. Feather wrote:
> The original suggestion was to keep abstracts around much longer and in
> many more places than full articles. Looking at Usenet, I suggest that
> using the first 40 lines as a default abstract would be useless.

No, that was not the original suggestion. The original suggestion was that
those who wanted to see more than 40 lines, could retrieve them, using an
URL supplied with the message. I do not understand why this would be
useless. Probably many recipients of long messages do not want to read more
than 40 lines.

If your experience is that the first 40 lines are not a good abstracting
method, one could consider more intelligent methods, for example the first
40 lines not preceded by "> ". But my experience is that the first 40
lines work perfecly well for me in e-mail, Usenet News may be different.

Caching with backup of full documents at a remote site is already common
in WWW, why could it not also be used in Usenet News?

MIME already has a format for this:

Multipart/alternative

	First part Message/external with an URL to the place where you
	can retrieve the whole article

	Second part the abbreviated message with an addendum that it
	has been abbreviated

------------------------------------------------------------------------
(Continue reading)

Permalink | Reply | 14 comments |
headers
Clive D.W. Feather | 9 May 1997 16:20

Re: Where Usenet News can be better

Jacob Palme said:
>> The original suggestion was to keep abstracts around much longer and in
>> many more places than full articles. Looking at Usenet, I suggest that
>> using the first 40 lines as a default abstract would be useless.
> No, that was not the original suggestion.
[...]
> If your experience is that the first 40 lines are not a good abstracting
> method, one could consider more intelligent methods, for example the first
> 40 lines not preceded by "> ". But my experience is that the first 40
> lines work perfecly well for me in e-mail, Usenet News may be different.
[...]

Sorry, I'm not objecting to the concept as a whole (at least, not in that
message, I wasn't). Just the "first 40 lines" bit. While you may be right
for email (and I think I'd agree), Usenet is a *very* different beastie.

Perhaps "first 40 lines beginning with a letter" might work.

--

-- 
Clive D.W. Feather     | Work:  <clive <at> demon.net>  | Tel: +44 181 371 1138
Director of            | Home:  <clive <at> davros.org> | Fax: +44 181 371 1037
  Software Development | Abuse: <clive <at> bofh.org>
Demon Internet Ltd.    |
Permalink | Reply | 13 comments |
headers
Greg Berigan | 9 May 1997 17:28
Favicon

Re: Where Usenet News can be better


CDWF>Sorry, I'm not objecting to the concept as a whole (at least, not in that
CDWF>message, I wasn't). Just the "first 40 lines" bit. While you may be right
CDWF>for email (and I think I'd agree), Usenet is a *very* different beastie.
CDWF>
CDWF>Perhaps "first 40 lines beginning with a letter" might work.

I don't think so.  The above quoting style is too common.

The software is in a position to know what was quoted material and what
is original.  It knows what the quoting string is.  Therefore it is in
the best position to know where the real additional text is located and
to extract it for summaries accordingly.

Alternatively, perhaps some way to communicate exactly what quoting
method is in use in an article could be done, although I'd prefer a rule
saying that ">" is the only acceptable quoted text delimiter string.

--

-- 
         _-<#)-=# http://cse.unl.edu/~gberigan/War-of-the-Worlds.html
     ___/___
  _-~_--<###)    "Take a look around you at the world we've come to know
<~c>' __--<       Does it seem to be much more than a crazy circus show
 \_--=____#)      Maybe from the madness something beautiful will grow...."
Permalink | Reply | 3 comments |

Gmane