From: John Stanley (stanley@peak.org)
Date: Fri Jun 01 2001 - 13:55:32 CDT
Peter Alfredsen (peteralf@fabel.dk):
>I am not using an archiving tool to do any archving.
Thus you clearly and unambiguously are not running an archive. How silly
of us to think otherwise.
>1. Breach of current practice - a newsserver
Archive. Archive header applies to archive. It's not rocket science.
>has never been required to
> be able to delete/make invisible[1] certain normal posts before.
We have never done it before, thus we can never do it. Please stop making
this ridiculous argument. Just imagine what news would be like were your
argument anything but nonsense. Gosh, there wouldn't be news, since we
hadn't done it before so we couldn't do it now. Hey, guys, shut them damn
nodes down, didn't you know, we have never done the Internet so we can't
do it now!
>2. Taking a law and making it a technical standard.
Nonsense. Taking a feature that authors clearly want and are using today
and including it in the standard which is supposed to document existing
practice. Whatever law you are whining about has nothing to do with it.
>3. The fact that it will make Many servers non-compliant with the
> standard,
The archive header deals with archives, not servers. Even so, this draft
will make ALL servers non-compliant the day it is published as a standard.
Look at all the new nonsense encoded in the Path header that they won't be
encoding. You want to talk about "sudden" and "new"?
>4. The fact that people actually will expect the Archive: no header to
> matter
Gosh, what an awful thing to have happen. Gee, why bother creating any
header, if we cannot expect them to actually mean something?
>[1] Apparently John Stanley has a problem wit hme calling this an
> "expire" process,
Since it is not an expire process, yes, I have a problem with you trying
to call it that. And with you whining about servers having to do things
they didn't have to do before.
> but that would probably be how it would be
> implemented (that Archive: no articles are put on a different expire
> time than other articles)
It isn't rocket science. Archive: no means you don't keep the article at
all. If you want, you can pretend that you've already got it (ala history)
or you don't get that group (ala Newsgroups) or the date is too old (ala
Date). But pretending that you've got to keep a copy and then go back
later to get rid of it is STUPID. Patently stupid. Archive admins who
treat it as an "expire" are acting stupidly.
>As has been reiterated many times, it will be practically
>impossible to find a definition of archiving agent, that does not
>collide with the definition of a server, as servers can act as archiving
>agents.
Yes, you've said it many times. That doesn't make it true. And it does not
mean that one cannot differentiate between a site's function as an archive
and its function as a server.
>Also, Archiving agents such as the archive I have put up at
>sunsite.dk is perfectly readable by many newsreaders,
So you admit that it is an archive now. Or are you admitting it here only
to make this one point, and you'll go back to claiming it is not an
archive when the mood strikes?
>as it uses a standard format to store the articles in.
Bzzzzt. Sorry, there is no "standard format" to store articles. Are you
sure you are participating the the working group you think you are?
>This is a "build it and they will come" argument. We can not suddenly
>make the Archive header a technical requirement,
Bzzzt. Thanks for playing. The Archive header is the standards format of
the X-No-Archive header, which has been in use for many years already.
There is no "sudden" about it.
>X-No-Archive was purely meant for the Deja Archive. Not for all the
>other archives of usenet that exist.
Existing practice contradicts you. Yes, Deja came up with it. No, people
do not think it applies only to Deja.
>The only two cases that I have seen
>an implementation of the X-No-Archive header is at Deja.com and
>remarq.com.
Bzzzzt. Thanks for playing. Every article I post has that header in it.
>and it is certainly
>not up to us to use imperatives from 2119 stronger than MAY, as no
>interoperability problems have been pointed out yet.
Bzzzt. Thanks for playing. This draft ignores RFC2119 when it wants to, so
doing it here is certainly not a problem.
>I can accept the
>Archive header if it is used with MAY or Ought.
You've previously said you will not honor this header at all. We are
expected to believe that you WILL honor it if it says Ought but won't if
it says MUST?
>The problem you are missing here, is that normal servers will fall under
>the definitions of an "archiving agent" that has been brought up so far.
Say it twice more, maybe it will become a fact.
>The problem for me, right now, is that my "Archive" doesn't fit into the
>definition of Archive you have, as it is offered by FTP,
First you claim that it is accessible by standard newsreaders, now you
claim it is FTP. What difference does it make what transport method the
articles are accessible via? Do you have some notion that USENET isn't
USENET if it ain't NNTP?
>I know of servers that has articles dating back 3-4 years,
>and as they are not Archives,
How do we know this?
>we are going to have servers that are suddenly also
>required (because it may cause interoperability problems) to remove, or
>make unavailable, articles.
Accepting this FOR THE MOMENT as true, so what? Big boo hoo. I'm simply
crying in my Mt. Dew over this unbearable condition being put on servers.
My God, they may actually be able to throw articles away AT THE AUTHORS
REQUEST and lengthen the expiration times on other groups. They may be
able to make MORE news available to their users without having to buy
bigger disks and without any ethical issues about removing an article
because someone else thinks it is spam.
Charles Lindsey (chl@clw.cs.man.ac.uk):
>I think it likely that a system that only keeps half a dozen groups will
>also not be publicly accessible, so in that case the question will not
>arise.
Outside any other issues, I disagree with that. When one group I was
interested in was created, I created a public archive of that group at the
same time. One group. I've since closed it, since people said that "Deja
did it better". Of course, my archive existed before Deja started and
exists after, but Deja is better anyway.
>Ok, I have pieced this together from the other proposed wordings. It is
>plain, it is simple, and it is what I really want -
And your wording creates a meaningless, useless header. If that is what
you want, be honest enough to argue that up front and not via pretense.
J.B. Moreno (planb@newsreaders.com):
>but it seems to me that the majority of people agree with me that we
>can't/shouldn't make it part of the protocol and advisory is the way to
>go, is this so?
No. It is existing practice and should be part of the standard and it is
not advisory, it is an instruction from the author.
Re: Creating expiring groups, Dennis SCP (dennis.scp@multiweb.nl):
>Users expect articles to be kept as long as newsserver storage
>capability can provide.
My local newsserver capability looks to be a week or so. How does my local
newsserver capability define anything? Or are we now agreed that anything
longer than a "week or so" is an archvie?
>There is no date based maximum expected retention by the users.
Really? Where did that Expires header go, then? I was sure that it
contained a date.
>Stop beating around the bush and just define a standard offset to the
>Date header when Archive is set to no. Say 30 days or 100 days.
What is so hard about this? NO means NO. It does not mean "in 30 days" or
"in 100 days" or "when you think you might get around to it".