* Tue, 31 Jul 2007 23:40:11 +0200
>
[]
> eventhough people often write only when they have something to complain
> about, I for once would like to congratulate Matti and David, our mail
> admins, for the wonderful job they've done with the spams lately. This
> month, I might have seen maybe one or two instead of perhaps 30 per day
> earlier. While it did not bother me that much earlier, I can say that
> the list is more pleasant to read every day, and I think that's great.
>
> Kudos guys !
>
> Willy
>
> PS: please do not start another one of those long useless threads from now on
>
Instead let me share an idea and see what will happen (if any).
The mailing list policy:
1. sent mails must have in-reply-to (rfc2821 SHOULD) with valid, i.e.
existing in archives, message-id;
2. otherwise sender gets response about creating new thread. Replying
to that message will enable (1).
I don't know how performance of ever increasing anti-spam
rules/software grow vs message-id lookup will be affected.
Gmane shows good performance on load, which is times bigger than LKML's.
News server requires unique ids, so...
Ah, if spammers will become smart and will reply with appropriate headers
set up, everything will fall apart.
Finally. What i see:
* no spam,
* big patch-bombs in one thread (easy to skip)
* more order, if useless/obsolete footers removed, but such info is
sent in (2).
* easy blacklisting if rules are not obeyed since day X
Oh, no, don't say we value bug reports. Reports must be organized in
other way, one of which i've described as development tracking system.
Such e-mail policy is unavoidable part of this it. E-mail is everything
it requires. Of course some html, css, javascript, perl guys will be
free to implement any web-face they like. But story if git-web have
something to learn instead.
--
-o--=O`C
#oo'L O
<___=E M
On 2007.09.02 02:36:18 +0200, Oleg Verych wrote:
> * Tue, 31 Jul 2007 23:40:11 +0200
> >
> []
> > eventhough people often write only when they have something to complain
> > about, I for once would like to congratulate Matti and David, our mail
> > admins, for the wonderful job they've done with the spams lately. This
> > month, I might have seen maybe one or two instead of perhaps 30 per day
> > earlier. While it did not bother me that much earlier, I can say that
> > the list is more pleasant to read every day, and I think that's great.
> >
> > Kudos guys !
> >
> > Willy
> >
> > PS: please do not start another one of those long useless threads from now on
> >
>
> Instead let me share an idea and see what will happen (if any).
> The mailing list policy:
>
> 1. sent mails must have in-reply-to (rfc2821 SHOULD) with valid, i.e.
> existing in archives, message-id;
>
> 2. otherwise sender gets response about creating new thread. Replying
> to that message will enable (1).
Ehrm, you want everyone who wants to start a new thread to:
- send an email
- await response from the mail server
- send the same email again as a reply to the first one
Did I get that right? If so, good luck running from anyone who doesn't
send his patch series as a bunch of replies to a 00/XX message.
Although it would likely be fun to see the next guy that sends a 600
emails patch bomb getting buried alive in "Please confirm that you want
to create a new thread" emails.
Bj?rn
On Sun, Sep 02, 2007 at 02:39:36AM +0200, Bj?rn Steinbrink wrote:
[]
>
> Ehrm, you want everyone who wants to start a new thread to:
>
> - send an email
> - await response from the mail server
> - send the same email again as a reply to the first one
No. Send a ticket request and then organize patch bomb with gotten ticket
in `references:' or `in-reply-to'. New discussion thread: be organized
have a pull of tickets, care about what you post, how you post, etc.
Those new posters will get current footer with more info -- no problem.
People in thread need no work at all. Linus, Andrew, Jeff, DaveM etc.
are likely to be in white list with no additional work also.
It's anti-spam thing, but maybe it can be kind of netiquette, which is
forgotten in modern XML Internet world. I didn't live in that time, but
that's my impression. In case if efficient algo/logic will be implemented
it will save CPU power, because all processing is done in header part of
the message and it's a token lookup. Switch on your imagination and
think, what can be done in another way, why not?
____
On Sun, Sep 02, 2007 at 03:23:53AM +0200, Oleg Verych wrote:
> On Sun, Sep 02, 2007 at 02:39:36AM +0200, Bj?rn Steinbrink wrote:
> []
> >
> > Ehrm, you want everyone who wants to start a new thread to:
> >
> > - send an email
> > - await response from the mail server
> > - send the same email again as a reply to the first one
>
> No. Send a ticket request and then organize patch bomb with gotten ticket
> in `references:' or `in-reply-to'. New discussion thread: be organized
> have a pull of tickets, care about what you post, how you post, etc.
>
> Those new posters will get current footer with more info -- no problem.
>
> People in thread need no work at all. Linus, Andrew, Jeff, DaveM etc.
> are likely to be in white list with no additional work also.
>
> It's anti-spam thing, but maybe it can be kind of netiquette, which is
> forgotten in modern XML Internet world. I didn't live in that time, but
> that's my impression. In case if efficient algo/logic will be implemented
> it will save CPU power, because all processing is done in header part of
> the message and it's a token lookup. Switch on your imagination and
> think, what can be done in another way, why not?
While probably efficient for a few months, it will be terribly boring
to use. I would most probably unsubscribe from then. The most efficient
system against spammers is not to annoy the users, but the spammers.
Otherwise they win.
Your method makes it trivial for them to request 1 ticket, then send 10000
mails which will be accepted without ever being filtered. It will result
in an increase of the number of spams and a great annoyance for the users.
Not for me, thanks :-/
Willy
On Sun, Sep 02, 2007 at 10:18:36AM +0200, Willy Tarreau wrote:
[]
>
> Your method
as i mentioned
> makes it trivial for them to request 1 ticket, then send 10000 mails
> which will be accepted without ever being filtered. It will result in
> an increase of the number of spams and a great annoyance for the users.
It's metter of experiment to know how much spammers today
* compose messages with non outloock default headers
* have *valid* reply address
** if it's valid and message to get ticket looks spammy as well as message
after ticket does, address gets blacklisted forever
* bother about published email address' policy
> Not for me, thanks :-/
You know, after requests like "we need more man power to handle mail
lists" i wonder in what century we are leaving XXI or IIX?
OTHO Gmane is an excellent example of applying NNTP structure, that is
very superior to mailing lists concept. But as i mentioned, i don't know
whole story about USENET vs Mailing lists. I just speculate, that latter
was established as an private entity, that groups of people can handle as
they need and want.
But NNTP isn't USENET, and only Gmane shows how it lets people like me to
handle tens of thousand emails in nearly 60 mailing lists (not only
reading up to date, but archives easily). And frankly, i just pissed off
by "6 day lkml archive, please", "Adrian, your email is bouncing, bla
bla...". I'd better see spam, then dumb things like that, see?
____