separating ruby-talk from comp.lang.ruby?

[Jason Creighton]

Hi --

Given that the gateway between ruby-talk and comp.lang.ruby is still
not working consistently, I'm wondering whether it's time to turn off
the gateway entirely and separate the two groups. It seems to me that
partial mirroring is worse than no mirroring.

Dennis (the maintainer) has tried to analyze and fix the problem, but
it appears to be downstream from his system somewhere. I guess it's
possible to investigate further, but at this point it may be pretty
hard, and meanwhile the unofficial split between the list and the
newsgroup continues to grow. I think it might be better to make it
official.

I don't know exactly how such a decision would be made, but I figured
the first step would be to see what people think of such a move (not
the idea of mirroring per se, but the idea of turning off mirroring in
light of the fact that it doesn't work). So.... ?

My 2 cents:

I think mirroring the newsgroup and the mailing list is great. Just
wonderful. But given that the mirroring isn't working very well, one of
the following should be done:

1. Fix the current setup. Obviously, this has been tried, so I'm not
holding out much hope on this one.

2. Pull the plug on the current setup and try some other mail-news
gateway software. (I haven't looked, but there's probably tons of them
floating around the net.)

So, if (1) isn't going to work for us, perhaps it would be best to make
the split offcial until we[1] can fix it.

[1] I mean, of course, anybody but me. I don't have a suitable machine
available, otherwise I might give it a shot.

Jason Creighton

 

[Yukihiro Matsumoto]

Hi,

In message "Re: separating ruby-talk from comp.lang.ruby?"

|Is there somewhere a full ml archive for download?
|(and a good idea to diff them)?

You can get them from <ftp://ftp.ruby-lang.org/ruby/ML/>.
Please don't tease our bandwidth too heavily, though.

matz.

 

[Tanu Kaskinen]

Is anybody aware of other services which does perfect mirroring (except NSA)?

No one has yet suggested Gmane, maybe there is a good reason for that...
I don't have any information about it's perfectness, but currently 5615
mailing lists are using Gmane. See http://gmane.org/ for more information.

The downsides I'm aware of:
* comp.lang.ruby would change its name
* Google wouldn't archive the newsgroup
* News users would have to add a new server in their reader in
order to read the group

 

[ts]

T> No one has yet suggested Gmane, maybe there is a good reason for that...

This ?

http://news.gmane.org/index.php?prefix=gmane.comp.lang.ruby

 

[Tanu Kaskinen]

T> No one has yet suggested Gmane, maybe there is a good reason for that...

This ?

http://news.gmane.org/index.php?prefix=gmane.comp.lang.ruby

Oh, I hadn't checked if this list was already mirrored via Gmane.

So gmane.comp.lang.ruby.general isn't the solution then?

 

[ts]

T> So gmane.comp.lang.ruby.general isn't the solution then?

I don't know but apparently c.l.r and ruby-talk are separated actually

 

[Tanu Kaskinen]

T> So gmane.comp.lang.ruby.general isn't the solution then?

I don't know but apparently c.l.r and ruby-talk are separated actually

I've got the impression that every post gets to the ruby-talk, and only
c.l.r would be incomplete.

Is anyone here using Gmane? Is it better than the official newsgroup?

 

[ts]

T> I've got the impression that every post gets to the ruby-talk, and only
T> c.l.r would be incomplete.

Well actually (since 3 or 4 hours) : I don't see on my news server the
message that I've mailed to ruby-talk, I don't see in my mailbox the
message that I've posted to c.l.r

 

[Gavin Sinclair]

Some people like myself don't have the time or inclination to be
subscribed to YAHVML (yet another high-volume mailing list), but would
like to participate and/or lurk on a semiregular basis. The newsgroup,
however imperfect, is perfect for this. A mailing list, archives or no,
is not. So my vote is keep the newsgroup!

The problem with semiregular lurkers is that they don't notice
Professor Black's repeated reminders that the newsgroup is not going
anywhere

Cheers,
Gavin

 

[Martin Pirker]

You can get them from <ftp://ftp.ruby-lang.org/ruby/ML/>.

got a full copy
a quick diff by counting message ids:

40419 in ml archive 60000-101400
39565 in my news archive since 20030101

adding both + sort + uniq gives 41533 unique ids

so the gateway isn't that bad, but there is only a 95% exchange

(errors: measurement is not day exact und in the ml archive
there is 83800- missing and 83700- is too small)


Martin

 

[David A. Black]

Hi --

got a full copy
a quick diff by counting message ids:

40419 in ml archive 60000-101400
39565 in my news archive since 20030101

adding both + sort + uniq gives 41533 unique ids

so the gateway isn't that bad, but there is only a 95% exchange

Over time, yes, but the recent problems are a lot more acute The
current situation seems to have started quite suddenly in late January
of this year. The last message of mine that made it from ruby-talk to
comp.lang.ruby was on January 24 (except for those where I initiate
threads, which do make it through).


David

 

[Robert Klemme]

Hi --



Over time, yes, but the recent problems are a lot more acute The
current situation seems to have started quite suddenly in late January
of this year. The last message of mine that made it from ruby-talk to
comp.lang.ruby was on January 24 (except for those where I initiate
threads, which do make it through).

Hm, this sounds like the problem could be header related. I guess, a
posting that starts a new thread has slightly different headers (namely
missing "references" and the like). Just a small difference but this could
be a hint...

Kind regards

robert

 

[David Alan Black]

Hi --

Hm, this sounds like the problem could be header related. I guess, a
posting that starts a new thread has slightly different headers (namely
missing "references" and the like). Just a small difference but this could
be a hint...

I sent Dennis comparative sets of headers a few months ago, when all
this started -- we weren't able to find anything, but please feel free
to have a look and see if you spot anything. I've just sent a new
dummy message to ruby-talk, which indeed got propagated to clr. For
recent examples of ones that didn't, see Joao's posts and my responses
to him, such as ruby-talk 101837.


David

 

[Martin Pirker]

Over time, yes, but the recent problems are a lot more acute The
current situation seems to have started quite suddenly in late January
of this year.

ok, rerun

ml:
91200-101400 (1.feb-25.may)
10201 files
grep -i ^message-id -> 10236 (?!?)
sort |uniq -> 9560

(same period)
8633 files -> 8633 message-ids
sort|uniq -> 8607

appending both files |sort |uniq -> 9586 unique message-ids

diff ml -> news 27 new lines
diff news -> ml 980 new lines


HTH,
Martin

 

[Robert Klemme]

Hi --



I sent Dennis comparative sets of headers a few months ago, when all
this started -- we weren't able to find anything, but please feel free
to have a look and see if you spot anything. I've just sent a new
dummy message to ruby-talk, which indeed got propagated to clr. For
recent examples of ones that didn't, see Joao's posts and my responses
to him, such as ruby-talk 101837.

Too many hops maybe? There are a lot "Received" lines:

Received: from kankan.nagaokaut.ac.jp (kankan.nagaokaut.ac.jp
[133.44.2.24])
by blade.nagaokaut.ac.jp (8.12.3/8.12.3/Debian-6.6) with ESMTP id
i4UNIOZu028628;
Mon, 31 May 2004 08:18:24 +0900
Received: from funfun.nagaokaut.ac.jp (funfun.nagaokaut.ac.jp
[133.44.2.201])
by kankan.nagaokaut.ac.jp (Postfix) with ESMTP
id 345CE58BC; Mon, 31 May 2004 08:18:24 +0900 (JST)
Received: from localhost (localhost.nagaokaut.ac.jp [127.0.0.1])
by funfun.nagaokaut.ac.jp (Postfix) with ESMTP
id 065A7F04846; Mon, 31 May 2004 08:18:25 +0900 (JST)
Received: from voscc.nagaokaut.ac.jp (voscc.nagaokaut.ac.jp
[133.44.1.100])
by funfun.nagaokaut.ac.jp (Postfix) with ESMTP
id 24E8BF04898; Mon, 31 May 2004 08:18:24 +0900 (JST)
Received: from helium.ruby-lang.org (helium.ruby-lang.org
[210.251.121.214])
by voscc.nagaokaut.ac.jp (Postfix) with ESMTP
id 7364763003F; Mon, 31 May 2004 08:18:23 +0900 (JST)
Received: from helium.ruby-lang.org (localhost [127.0.0.1])
by helium.ruby-lang.org (Postfix) with ESMTP
id 3C48E427004; Mon, 31 May 2004 08:18:21 +0900 (JST)
Received: from wobblini.net (wobblini.net [65.200.24.105])
by helium.ruby-lang.org (Postfix) with SMTP id CAF15426FFB
for <[email protected]>; Mon, 31 May 2004 08:18:15 +0900
(JST)
Received: (qmail 9922 invoked by uid 500); 30 May 2004 23:18:13 -0000
Received: from localhost ([email protected])
by localhost with SMTP; 30 May 2004 23:18:13 -0000

(Note: I substituted tabs by spaces for reader independend formatting)

see http://blade.nagaokaut.ac.jp/ruby/ruby-talk/101837

Regards

robert

 

[daz]

From: Robert Klemme

Hi --

I sent Dennis comparative sets of headers a few months ago, when all
this started -- we weren't able to find anything, but please feel free
to have a look and see if you spot anything. I've just sent a new
dummy message to ruby-talk, which indeed got propagated to clr. For
recent examples of ones that didn't, see Joao's posts and my responses
to him, such as ruby-talk 101837.

Too many hops maybe? There are a lot "Received" lines:

Received: from kankan.nagaokaut.ac.jp (kankan.nagaokaut.ac.jp
[...]

============================================================
Anyone looking at header differences, please save your time.
============================================================

There's much processing at nagaokaut.ac.jp and helium.ruby-lang.org
such as spam checking, virus scanning, authentication, MIME type
conversion, character code conversion and stuff. At the end, it
seems that list members take delivery of all messages posted to the list.
There's a question in my mind about whether the gateway is treated
as an equal member since February, though.


daz

 

[daz]

============================================================
Anyone looking at header differences, please save your time.
============================================================

====================================================================
Anyone looking for a remote cave to hide in, please stand behind me.
====================================================================

Still, if anyone can explain why I can post with "In-Reply-To:"
but without "References:" in the headers but Dennis can't, please
post a reference so that I can check it in 20 years time (if I
choose to re-emerge).


#====================================================================================================
Newsgroups: comp.lang.ruby
X-received-from: This message has been automatically forwarded from
the ruby-talk mailing list by daz (testing).
Date: Tue, 1 Jun 2004 00:00:37 +0900
From: "David A. Black" <[email protected]>
Subject: Re: separating ruby-talk from comp.lang.ruby?
Message-Id: <Pine.LNX.4.44.0405310800060.20545-100000@wobblini>
In-Reply-To: <[email protected]>
X-ML-Name: ruby-talk
X-Mail-Count: 101890
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-ruby-talk: <Pine.LNX.4.44.0405310800060.20545-100000@wobblini>
X-rubymirror: yes

Sorry everyone, testing again.... small meaningless tweak to
headers....

--
David A. Black
(e-mail address removed)


#====================================================================================================



daz

 

[Robert Klemme]

====================================================================
Anyone looking for a remote cave to hide in, please stand behind me.
====================================================================

Still, if anyone can explain why I can post with "In-Reply-To:"
but without "References:" in the headers but Dennis can't, please
post a reference so that I can check it in 20 years time (if I
choose to re-emerge).

I think Dennis meant to say that this doesn't work for mail2
Dennis at Tue, 01 Jun 2004 17:09:27 +0200:
Subject: Tue, 01 Jun 2004 17:09:27 +0200

I'm not absolutely convinced that the problem is solved 100%. The bug
we solved today was that messages with an 'In-Reply-To'-header, but
without a 'References'-header were rejected by the NNTP host, because
it is a violation of the corresponding RFC. The solution was as simple
as the problem - adding a 'References'-header with the same msg-id as
the 'In-Reply-To'-header solved at least _this_ problem.

As far as I can see, you post via news. Is that correct?

Regards

robert

 

[daz]

Still, if anyone can explain why I can post with "In-Reply-To:"
but without "References:" in the headers but Dennis can't, please
post a reference [...]

I think Dennis meant to say that this doesn't work for mail2
Dennis at Tue, 01 Jun 2004 17:09:27 +0200:
Subject: Tue, 01 Jun 2004 17:09:27 +0200

I'm not absolutely convinced that the problem is solved 100%. The bug
we solved today was that messages with an 'In-Reply-To'-header, but
without a 'References'-header were rejected by the NNTP host, because
it is a violation of the corresponding RFC. The solution was as simple
as the problem - adding a 'References'-header with the same msg-id as
the 'In-Reply-To'-header solved at least _this_ problem.

As far as I can see, you post via news. Is that correct?

<Late response>

Normally, but that was sent using the gateway software running
on my machine. I subscribed to ruby-talk, chose that message,
and pushed it through.

As I said elsewhere, if Dennis's news service saw it as invalid, it
should be ignored then passed around Usenet until it reaches Giganews
and Google and all the others which _don't_ think it's invalid and it
should appear normally for users of those other services.

I think that his error messages are coming from his NNTP host between
rubygate and Usenet and the posts are not getting to Usenet.
The RFC to which he refers must be describing the standards imposed
by his NNTP host, not Usenet.

My evidence:
http://groups.google.com/groups?selm=Pine.LNX.4.44.0405310800060.20545-100000@wobblini&output=gplain

.... confirms that Usenet cares about http://www.w3.org/Protocols/rfc1036/rfc1036.html

<extract>
2.2. Optional Headers
^^^^^^^^
2.2.5. References
[...] It is required for all follow-up messages, [...]
</extract>

(Optionally required.)
That post of David's didn't have References but, even if it was invalid,
it's on Google and it wasn't before.


daz

 

[Robert Klemme]

Still, if anyone can explain why I can post with "In-Reply-To:"
but without "References:" in the headers but Dennis can't, please
post a reference [...]

I think Dennis meant to say that this doesn't work for mail2
Dennis at Tue, 01 Jun 2004 17:09:27 +0200:
Subject: Tue, 01 Jun 2004 17:09:27 +0200

I'm not absolutely convinced that the problem is solved 100%. The bug
we solved today was that messages with an 'In-Reply-To'-header, but
without a 'References'-header were rejected by the NNTP host, because
it is a violation of the corresponding RFC. The solution was as simple
as the problem - adding a 'References'-header with the same msg-id as
the 'In-Reply-To'-header solved at least _this_ problem.

As far as I can see, you post via news. Is that correct?

<Late response>

Normally, but that was sent using the gateway software running
on my machine. I subscribed to ruby-talk, chose that message,
and pushed it through.

As I said elsewhere, if Dennis's news service saw it as invalid, it
should be ignored then passed around Usenet until it reaches Giganews
and Google and all the others which _don't_ think it's invalid and it
should appear normally for users of those other services.

I think that his error messages are coming from his NNTP host between
rubygate and Usenet and the posts are not getting to Usenet.
The RFC to which he refers must be describing the standards imposed
by his NNTP host, not Usenet.

I think I need a clarification here: my understanding so far was, that
"Usenet" denotes the system of distributed news groups, while "NNTP" is
the protocol used for the transmission of news articles between servers.
Are there additional protocols used for news exchange?

Thx

robert