Re: [Gossip] Use of forwarded e-mail to circumvent confirmationrequirement.

2003-10-29 Thread Jeff Breidenbach

>So, would it be possible to use the "To:" address for the name of the
>mailing list rather than the "Delivered-To:"?  Pretty please?

Ok, so here is what I think was happening. The "subscribe
confirmation" message to the wedi-* lists got archived under
wedi-mirror.  This is actually a reasonable thing to do. From
Mail-Archive's perspective, it really does look like wedi-mirror 
was the list name.

Now that wedi-mirror archive was started, incoming messages matched
two lists, wedi-mirror and wedi-. As soon as that happens,
Mail-Archive takes a closer look at headers and tries to disambiguate.
It sees wedi-mirror in the Delivered-To: and says "ah ha! We've
disambiguated." And files under wedi-mirror

I've gone ahead and moved the wedi-mirror archives aside. Now that
there is no disambiguation step required, mail is archiving ok.  We
match just one list, wedi-. But it is very fragile, and the
second some message creates a wedi-mirror archive things will go south
again.

Bottom line - I've got things kinda working, but it could break again
at any time. This tunneling though a mail alias business does not play
well with the sorting heuristics. And I'm not going to optimize the
sorting heuristics for tunneling (especially if it hurts the
non-tunneling cases). Nothing against wedi, but I often associate
tunneling with abuse. Such as lists trying to get around the
YahooGroups ban or someone trying to covertly archive a mailing list.
That's an additional motivation not to spend effort support tunneling.

In an unrelated note, Mail-Archive survived today's slashdotting just
fine. We served nearly a million hits today. After I upped the number
of allowed simultaneous HTTP connections to about 550, there was no
problem keeping up with the load.

-Jeff

___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip


Re: [Gossip] CSS and RSS

2003-10-29 Thread Dror Matalon
Hi,


> I am going to study how mailbucket.org and get back to you.  
> RSS is quite confusing atm. Different versions et al.  
> I am going to study how mailbucket.org and get back to you.  
> RSS is quite confusing atm. Different versions et al.

We offer a free RSS aggregator that should make it quite easy to do this
at www.fastbuzz.com. 

You can choose either web or email delivery, and it's quite easy to
add mail-archive lists to your rss channels.

This is how I follow some of the lists that I'm interested in.

Regards,

Dror


-- 
Dror Matalon
Zapatec Inc 
1700 MLK Way
Berkeley, CA 94709
http://www.zapatec.com

___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip


[Gossip] Three questions

2003-10-29 Thread Kristian Köhntopp
Question the first: How does mail-archive.org act when a reply to one mail is 
sent very much later, say one to five years? The message will be a reply to 
the original message, with In-Reply-To headers and everything else set 
correctly, and will be part of the mailing-list that mail-archive.org is 
archiving.

Context: Under german law everybody mentioned in a public message has the 
right to post his disagreeing view in the context of the original message. 
That means that for example newspapers that print news about Person x, who 
has done such and so, must print the statement of Person x regarding the same 
matter ("Contrary to the claims of the paper, I haven't done..."), and do so 
unedited. There are some limits  to the size of that statement and so on, but 
essentially everybody has the right to deny and correct, and the right hat 
their correction is being printed unedited. The newspaper is free to add 
their view of things after that unedited statement, for example that they 
still uphold their original opinion based on the facts reported in the 
original article.

This law also applies to mailing-lists, USENET news and so on. Some people on 
debate asked how that law could be applied to mailing-lists, and I stated 
that simply replying to the original message will probably satisfy the 
requirements of that law. This is, because the reply message will be properly 
linked into the context of the original message due to In-Reply-To and/or 
References headers. Then I asked myself what an archive might do if there is 
more than ~1 month or so between the original statement and the reply. Will 
the archive still thread that message or will the thread be broken?


Even more context: [EMAIL PROTECTED] is a german language mailing-list 
dealing with information society politics stuff which has a public archive on 
http://www.fitug.de. Because in the past that archive has been the target of 
history correctors that want things they said in the past unsaid or changed, 
I started to forward my copy of debate to mail-archive.org to create a public 
archive in another jurisdiction. My hope is that an archive being out of 
german jurisdiction will discourage these people.

I chose mail-archive.org because it is a non-commercial community project that 
seemed trustworthy and stable, and has a sensible spam policy (mail addresses 
removed, but reply-by-mailclient instead of webform still possible). I also 
chose mail-archive.org because of their one-word deletion policy.

While I am of the opinion that one should not be able to unsay things one said 
in a public forum (that would be tampering with history), one should be able 
to amend or correct that things in later messages, and it would be of much 
value, if these corrections would be presented in proper context as an 
addendum to the original thread.


Question the second: Is it possible for a user of mail-archive.org to 
determine which particular subscriber of a mailing-list is forwarding the 
list to mail-archive.org? 

From a quick look at the HTML it seems that it isn't, which I think is good.

Also, what is the policy of mail-archive.org regarding inquiries in that 
matter? That is, if asked, under what circumstances will mail-archive.org 
produce original headers of the messages sent to it, and offer the name of 
the submitter, that is, of the source that is feeding the archive?


Question the third: Assuming that there are two subscribers to 
[EMAIL PROTECTED] that forward their messages to mail-archive.org, will 
the archive detect the duplicate submissions and drop the duplicate copies?

This is ATM a hyothetical question. I don't want to burden the archive with 
doubled traffic. It may be handy to know, though, in case the feed must be 
handed over from one person to another, because if there is no duplicate 
weedout, the handover is more complicated.

Many thanks,
Kristian


___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip


Re: [Gossip] Three questions

2003-10-29 Thread Earl Hood
(Note, I do not run mail-archive.com, but I familiar with some its
 software)

On October 29, 2003 at 19:03, Kristian =?iso-8859-15?q?K=F6hntopp?= wrote:

> Question the first: How does mail-archive.org act when a reply to one 
> mail is 
> sent very much later, say one to five years? The message will be a reply 
> to 
> the original message, with In-Reply-To headers and everything else set 
> correctly, and will be part of the mailing-list that mail-archive.org is 
> archiving.
> 
> Context: Under german law everybody mentioned in a public message has the 
> right to post his disagreeing view in the context of the original 
> message. 
[snip]

If you are talking one to five years, there will probably not be an
explicitly link, unless list traffic is very light.  Your window of
explicit linking on mail-archive is the last 3000 messages (Jeff can
correct me if I'm wrong).

The original message will still be on the archive, but one must search
for it.  Now, if some form of message-id linking was performed,
something more direct can be done, but such a thing would require
changes to the underlying software.

As for the German law, it has no jurisdiction with the U.S., where
mail-archive is located.  And, IMHO, the German law you refer to
should not have been passed.

But in general, it is possible to have a mail archive system provide
explicit linking, despite the time between message replies, but
mail-archive is limited in this regard (as noted above).

> This law also applies to mailing-lists, USENET news and so on. Some 
> people on 
> debate asked how that law could be applied to mailing-lists, and I stated 
> that simply replying to the original message will probably satisfy the 
> requirements of that law. This is, because the reply message will be 
> properly 
> linked into the context of the original message due to In-Reply-To and/or 
> References headers. Then I asked myself what an archive might do if there 
> is 
> more than ~1 month or so between the original statement and the reply. 
> Will 
> the archive still thread that message or will the thread be broken?

You make the assumption that MUAs properly define in-reply-to and
references headers.  Outhouse ... excuse me ... Outlook fails miserably
in this regard.

Therefore, you assume that MUA software will play by certain rules,
and that is not always the case.

> Question the second: Is it possible for a user of mail-archive.org to 
> determine which particular subscriber of a mailing-list is forwarding the 
> list to mail-archive.org? 
> 
> From a quick look at the HTML it seems that it isn't, which I think is 
> good.

You are correct wrt the HTML.

> Also, what is the policy of mail-archive.org regarding inquiries in that 
> matter? That is, if asked, under what circumstances will mail-archive.org 
> produce original headers of the messages sent to it, and offer the name 
> of 
> the submitter, that is, of the source that is feeding the archive?

Jeff can only answer this one.  My guess is Jeff will refuse such
requests unless a subpoena is involved.

> Question the third: Assuming that there are two subscribers to 
> [EMAIL PROTECTED] that forward their messages to mail-archive.org, 
> will 
> the archive detect the duplicate submissions and drop the duplicate 
> copies?

For one software component that mail-archive uses, it will ignore
messages that have the same message-id to an existing archived one
(up to the 3000 message size window Jeff has specified).  However,
I do not know about the details of any other dup checks done before
a mesasge reaches this component.

--ewh

___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip


Re: [Gossip] Three questions

2003-10-29 Thread Earl Hood
On October 29, 2003 at 21:55, Kristian =?iso-8859-15?q?K=F6hntopp?= wrote:

> > And, IMHO, the German law you refer to should not have been 
> > passed.
> 
> I understand from a previous discussion on SlashDot that the concepts behind 
> this particular law are very foreign to people in the US.

I understand the reasoning of the German law, I just do not agree
with it.  The attentions are well meaning, but as the saying go,
"the road to hell is paved with good attentions."  I will skip my
reasons why I do not agree with law since I am sure it has been
debated to death on Slashdot.

> > You make the assumption that MUAs properly define in-reply-to and
> > references headers.  Outhouse ... excuse me ... Outlook fails miserably
> > in this regard.
> 
> In this particular application I think it is safe to assume that not just any
>  
> client is being used, but that the author prepares that statement using a 
> client that generates proper header, or even massages the message headers 
> manually in order to insure proper linking.

I answered your questions in a general context.  If your usage context
can guarantee proper headers are defined, then linking should work
as you expect.

> Since the vast majority of these types of statements are being prepared withi
> n 
> few days of the issueing of the original statement, a 3000 messages window is
>  
> just fine. If it isn't, that's tough luck...

But, ignoring jurisdiction, would you still have problems with conforming
to German law?  Do you care?

> > > Question the third: Assuming that there are two subscribers to
> > > [EMAIL PROTECTED] that forward their messages to mail-archive.org,
> > > will
> > > the archive detect the duplicate submissions and drop the duplicate
> > > copies?
> >
> > For one software component that mail-archive uses, it will ignore
> > messages that have the same message-id to an existing archived one
> > (up to the 3000 message size window Jeff has specified).
> 
> So a handover will probably be painless.

I'm unsure about what you exactly mean by "handover".  The way
mail-archive works is that you subscribe [EMAIL PROTECTED] to
the list you want archived.  Mail-archive can care less who admins
the lists and if admins change.  Now, if the list address changes,
you will need to be careful since it could case the creation
of a new archive based on the new list address.  Have a look at
.

--ewh

___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip


Re: [Gossip] CSS and RSS

2003-10-29 Thread Dror Matalon
Hi,

I've been playing with this a bit more. So now we provide a way to see
the list of items without logging in first.

So for instance for this list, the gossip list, you can go to 
http://www.fastbuzz.com/items/preview_items.jsp?channel=71296
and always see the latest items on the list.

So it can be an easy way for those interested to show the latest items
from a mailing list on their web pages.

Regards,

Dror

On Mon, Oct 27, 2003 at 11:14:44PM -0800, Dror Matalon wrote:
> Hi,
> 
> 
> > I am going to study how mailbucket.org and get back to you.  
> > RSS is quite confusing atm. Different versions et al.  
> > I am going to study how mailbucket.org and get back to you.  
> > RSS is quite confusing atm. Different versions et al.
> 
> We offer a free RSS aggregator that should make it quite easy to do this
> at www.fastbuzz.com. 
> 
> You can choose either web or email delivery, and it's quite easy to
> add mail-archive lists to your rss channels.
> 
> This is how I follow some of the lists that I'm interested in.
> 
> Regards,
> 
> Dror
> 
> 
> -- 
> Dror Matalon
> Zapatec Inc 
> 1700 MLK Way
> Berkeley, CA 94709
> http://www.zapatec.com
> 
> ___
> Gossip mailing list
> [EMAIL PROTECTED]
> http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip

-- 
Dror Matalon
Zapatec Inc 
1700 MLK Way
Berkeley, CA 94709
http://www.zapatec.com

___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip


Re: [Gossip] Three questions

2003-10-29 Thread Kristian Köhntopp
On Wednesday 29 October 2003 20:04, Earl Hood wrote:
> As for the German law, it has no jurisdiction with the U.S., where
> mail-archive is located.  

That is one particular reason why I chose mail-archive.org as a destination 
site for the mailing list as opposed to any site within the same jurisdiction 
the original list is in - makes tampering with archives harder for anyone 
involved.

> And, IMHO, the German law you refer to should not have been 
> passed.

I understand from a previous discussion on SlashDot that the concepts behind 
this particular law are very foreign to people in the US.

> You make the assumption that MUAs properly define in-reply-to and
> references headers.  Outhouse ... excuse me ... Outlook fails miserably
> in this regard.

In this particular application I think it is safe to assume that not just any 
client is being used, but that the author prepares that statement using a 
client that generates proper header, or even massages the message headers 
manually in order to insure proper linking.

Since the vast majority of these types of statements are being prepared within 
few days of the issueing of the original statement, a 3000 messages window is 
just fine. If it isn't, that's tough luck...

> > Also, what is the policy of mail-archive.org regarding inquiries in that
> > matter? That is, if asked, under what circumstances will mail-archive.org
> > produce original headers of the messages sent to it, and offer the name
> > of
> > the submitter, that is, of the source that is feeding the archive?
>
> Jeff can only answer this one.  My guess is Jeff will refuse such
> requests unless a subpoena is involved.

That would be pretty much the policy I am looking for.

> > Question the third: Assuming that there are two subscribers to
> > [EMAIL PROTECTED] that forward their messages to mail-archive.org,
> > will
> > the archive detect the duplicate submissions and drop the duplicate
> > copies?
>
> For one software component that mail-archive uses, it will ignore
> messages that have the same message-id to an existing archived one
> (up to the 3000 message size window Jeff has specified).

So a handover will probably be painless.

Many thanks for your helpful answers,
Kristian


___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip


[Gossip] Three questions

2003-10-29 Thread Jeff Breidenbach

>Question the first: How does mail-archive.org act when a reply to one
>mail is sent very much later, say one to five years? The message will
>be a reply to the original message, with In-Reply-To headers and
>everything else set correctly, and will be part of the mailing-list
>that mail-archive.org is archiving.

If less than N messages have been archived in the 5 year period, then
the reply will be threaded correctly on the thread index.  If more
than N messages have been archived in the 5 year period, there will be
no record of the original message in the thread index. The reply will
start a new thread.  Both messages will always be available via the
search engine interface, and N is roughly 3000.

>[...] Then I asked myself what an archive might do if there is more
>than ~1 month or so between the original statement and the
>reply. Will the archive still thread that message or will the thread
>be broken?

For Mail-Archive, the thread index (and chronological index) only
exists for the most recent N messages. As long as you are within
N, it should thread ok.

>I started to forward my copy of debate to mail-archive.org to create a
>public archive in another jurisdiction.

I would be much happier if the list administrator is the one who
initiates the archival process for three reasons. First, forwarding
(also known as tunneling) a list can cause technical problems with
Mail-Archive's sorting engine. Second, I don't interact with list
endusers if at all possible. I usually tell people to talk to their
list administrator. That's works a lot better when the list admin is
in the loop. Finally, Mail-Archive is for public lists desiring an
archive - it's not for archiving lists against the list admin's
wishes.

>Question the second: Is it possible for a user of mail-archive.org to 
>determine which particular subscriber of a mailing-list is forwarding the 
>list to mail-archive.org? 

If the forwarding messes up Mail-Archive's sorting engine, such as
what happened with WEDI, then it might be tracable. Otherwise I don't
think so.

>Also, what is the policy of mail-archive.org regarding inquiries in that 
>matter? That is, if asked, under what circumstances will mail-archive.org 
>produce original headers of the messages sent to it, and offer the name of 
>the submitter, that is, of the source that is feeding the archive?

Case by case. It hasn't really come up much. I have worked in the past
with a list administrator to resolve an abuse case, where mail was
being tunnelled to Mail-Archive against their list's policy. Also, any
header (or other) information ordered in a valid subpoena will be
disclosed.

>Question the third: Assuming that there are two subscribers to 
>[EMAIL PROTECTED] that forward their messages to mail-archive.org, will 
>the archive detect the duplicate submissions and drop the duplicate copies?

Depends on what the Message-ID is. If it is identical, I think we'll
drop the duplicate. At worst, we will store duplicate messages,
possible in the wrong place due to the tunneling messing things up.

Bottom line, please don't tunnel. Have your list admin add
[EMAIL PROTECTED] to the subscription list. If they are
unwilling, then please don't use Mail-Archive. Depending on how
much problems tunneling causes, I may move from discouraging to
explicitly banning them by policy in the future.

-Jeff

___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip


Re: [Gossip] Three questions

2003-10-29 Thread Jeff Breidenbach

Looks like Earl beat me to many of the same answers.  Mail-Archive
decisions and policies are made using my own (hopefully good)
judgement and my understanding of applicable US law.

-Jeff

___
Gossip mailing list
[EMAIL PROTECTED]
http://www.mail-archive.com/cgi-bin/mailman/listinfo/gossip