Separating ruby-talk from comp.lang.ruby?

Here, here, hear, hear. No NNTP here.

Drew

-> -----Original Message-----
-> From: Sven Schott [mailto:sven_schott@compnow.com.au]
-> Sent: Wednesday, May 26, 2004 6:52 PM
-> To: ruby-talk@ruby-lang.org
-> Subject: Re: separating ruby-talk from comp.lang.ruby?
->
->
-> Hear ,hear. I’m on the mailing list because of the those
-> very reasons.
-> I access everything through work.
->
-> Sven

···

-> On 26/05/2004, at 4:06 AM, Carl Youngblood wrote:
->
-> > Phil Tomson wrote:
-> >
-> >> I have always preferred reading comp.lang.ruby over subscribing to
-> >> ruby-talk (too much mail). I really hope we can keep the two
-> >> connected.
-> >>
-> > One other important thing to keep in mind is that some of us are
-> > forced to use ruby-talk because we don’t have access to a
-> news server
-> > (other than through google, which is dog slow). Even
-> though I prefer
-> > NNTP, I can’t use it at work.
-> >
-> > Carl
->
->

Hi –

Here, here, hear, hear. No NNTP here.

OK… but let me clarify: this isn’t about the question of which is
inherently better or more convenient, a newsgroup or a mailing list.
It’s about the specific situation we’re in, which is that we’ve got
two forums that try to mirror each other but don’t. If the gateway
were to be turned off, there would presumably still be a mailing list
and a newsgroup.

David

···

On Fri, 28 May 2004, Mills Thomas (app1tam) wrote:


David A. Black
dblack@wobblini.net

I very much regret that this discussion is happening.

I agree that it is unfortunate that some messages don’t make it between
the two. But the fact that some messages don’t get through doesn’t
seem like justification to change things such that 50% of messages no
longer get through (in either direction). I no longer read Usenet: I
don’t want to lose my Ruby friends who post using nntp because the link
gets cut.

It seems to me that we’re in a fairly string position here. My
understanding is that some of the problems are actually repeatable. In
particular DAB-initiated threads (I believe) are often not mirrored.

This situation seems amenable to experiment and debugging. David posts
a message, and a day later a Ruby script goes out to a list of news
servers and sees if it has arrived. If it arrives at all of them
(including the server used by the gateway) then the problem lies in the
gateway software or the mailing list. If it doesn’t, then the problem
might lie in David’s news host.

Similarly, if there are repeatable drops from the mail->news side, we
can test these too.

This is is bug, not a catastrophe. Let’s track it down rationally and
squash it.

Cheers

Dave

···

On May 27, 2004, at 15:06, David A. Black wrote:

OK… but let me clarify: this isn’t about the question of which is
inherently better or more convenient, a newsgroup or a mailing list.
It’s about the specific situation we’re in, which is that we’ve got
two forums that try to mirror each other but don’t. If the gateway
were to be turned off, there would presumably still be a mailing list
and a newsgroup.

Hi –

It seems to me that we’re in a fairly string position here. My
understanding is that some of the problems are actually repeatable. In
particular DAB-initiated threads (I believe) are often not mirrored.

It’s kind of the other way around: if I send a message that initiates
a thread, that one gets through (from mail to news). But if I reply
to anything on the mailing list, it doesn’t show up on news.

This situation seems amenable to experiment and debugging. David posts
a message, and a day later a Ruby script goes out to a list of news
servers and sees if it has arrived. If it arrives at all of them
(including the server used by the gateway) then the problem lies in the
gateway software or the mailing list. If it doesn’t, then the problem
might lie in David’s news host.

Similarly, if there are repeatable drops from the mail->news side, we
can test these too.

The bulk of the problem seems to be in this direction, judging by
comparing the list with what makes it to my news server and Google.

This is is bug, not a catastrophe. Let’s track it down rationally and
squash it.

I agree entirely, in principle – as I’ve said, my first choice is
definitely for there to be a working gateway, and my suggestion to
turn it off stems only from feeling utterly stymied after months of
discussions about its not working and no clues as to why.

If it’s helpful, I can certainly make available the full headers of (a
large number of :slight_smile: messages that were sent to the list but didn’t
make it to Usenet. I’ve sent some of these to Dennis, but my
understanding is there’s nothing he or his NNTP admins can spot.

(Paging Dennis…)

David

···

On Fri, 28 May 2004, Dave Thomas wrote:


David A. Black
dblack@wobblini.net

Hi –

[Note: In response to Dave’s comments I have change the subject
line to something a little more positive :-]

OK, here’s what Dennis most recently wrote me, summarizing what his
NNTP admins told him (pertaining to the mail-> news direction):

  • Tracing messages along their Path throughout the NNTP universe is
    practically impossible
  • There are NNTP hosts which reject certain messages
  • It is not possible to see if a certain message got rejected unless the
    host who rejected it is your direct up-/downstream
  • None of our up-/downstreams ever rejected messages from the
    Ruby-lang/c.l.r-gateway

You can see the roots of my pessimism :slight_smile: but I put this out there in
case it suggests anything to anyone, or a starting point for applying
Dave’s testing scenario to the mail->news direction.

David

···


David A. Black
dblack@wobblini.net

Hi –

···

On Fri, 28 May 2004, David A. Black wrote:

Hi –

On Fri, 28 May 2004, Dave Thomas wrote:

Similarly, if there are repeatable drops from the mail->news side, we
can test these too.

The bulk of the problem seems to be in this direction, judging by
comparing the list with what makes it to my news server and Google.

Update: I’m now seeing a lot of news posts that are not making it to
the mail side. I’d thought it was more the other direction, but it
seems to be both.

David


David A. Black
dblack@wobblini.net

Update: I'm now seeing a lot of news posts that are not making it to
the mail side. I'd thought it was more the other direction, but it
seems to be both.

It's broken since 26/05/2004 -- 16h00 (approx.)

Guy Decoux

So, see if these missing articles are making it to the news server used
by the gateway. If so, then the fault lies with the gateway software,
and it can be fixed. If not, then the fault lies with Usenet, and is
somewhat intractable.

Cheers

Dave

···

On May 28, 2004, at 7:55, David A. Black wrote:

Update: I’m now seeing a lot of news posts that are not making it to
the mail side. I’d thought it was more the other direction, but it
seems to be both.

David A. Black wrote:

Update: I’m now seeing a lot of news posts that are not making it to
the mail side. I’d thought it was more the other direction, but it
seems to be both.

For those of us who are interested in the matter, but aren’t aware of
how things work, can someone give a quick summary of how the mirroring
is supposed to work?

Some questions I have:

  • Is it a ‘fire and forget’ sort-of system – messages are sent to the
    NNTP server and if that doesn’t receive error messages it is assumed to
    have worked? Or does the software keep track of message IDs and somehow
    check to see that something has happened with them?

  • How are messages mirrored from the newsgroup to the mailing list? Is
    it just something like “Add all new messages as of this date/time to the
    mailing list”, or does it keep track of message IDs in some way?

  • Is there any ‘retry’ mechanism at all?

  • You mentioned that if the message isn’t being copied by a
    not-directly-upstream server, then there’s nothing that can be done to
    trace it. It seems like there should be. If you can confirm that the
    gateway is working properly with the immediate NNTP server, find out
    what its neighbors are and ask the admins there to verify that the
    messages were transferred correctly, then if so, plead with them to
    check with their neighbors. I would suspect that if everything is
    properly mirrored in all those places, then it will get really hard to fix.

I don’t know much about NNTP, only the basics. I don’t really
understand the network topology, or where our main node is on that
network. I also don’t know what would prevent a message from being
mirrored. Is the way that our messages are added to the NNTP server the
same as the way that a standard network client does it? Are there extra
headers or anything that might make an NNTP server upset?

Also, maybe the way the mailing list is set up can be changed. I
imagine right now that the way things work is:

Emails sent to the mailing list are sent out to all members,
one of those members is the gateway program, it sends them out
to the newsgroup. The newsgroup looks for new messages and when
it finds them, it emails them to the mailing list.

Maybe instead, we can have the mailing list be nothing but something
that copies messages to the newsgroup, and then have a secondary process
that takes everything from the newsgroup and sends it out to the mailing
list members. That way, instead of having two different lists that have
to be synchronized, we have only 1 (the newsgroup) that has to be
emailed out.

Any thoughts?

Ben Giddings

Hi –

Update: I’m now seeing a lot of news posts that are not making it to
the mail side. I’d thought it was more the other direction, but it
seems to be both.

So, see if these missing articles are making it to the news server used
by the gateway. If so, then the fault lies with the gateway software,
and it can be fixed. If not, then the fault lies with Usenet, and is
somewhat intractable.

Of course, right after posting that, I got an influx of about 30+
messages from the news side to the mailing list :slight_smile:

Anyway, I’m trying to get in touch with Dennis. My impression from
what he’s said over the past couple of months is that we’d already
reached the intractable problem phase (which is what led to my
original post in this thread), but I’ve asked him to check the log
files and the server content to get an up-to-date sense of what’s
happening (and not happening). I may also do some tests, just to have
a current dossier of tested scenarios.

On a meta-note, to everyone reading: I seem to have been misunderstood
rather widely on this matter. Please understand that I am NOT against
a news/mail mirror; I am frustrated by the illusion of one, so that I
don’t know whom I’m addressing, whether I’ve really answered a
question, whether I’m participating in a discussion or not, etc.

Moreover, my original post was not made on an impulse; I didn’t raise
the question of the gateway until I’d been communicating with Dennis,
the gateway maintainer, for 2.5 months, and been advised by him that
he and his NNTP admins had checked everything they could think of. It
may turn out that they missed something and that it’s fixable, and
that would be great. But either way, please do me the kindness of
believing that I actually put some thought and effort (and even
optimism) into the problem in the weeks and months before posting that
original message.

David

···

On Fri, 28 May 2004, Dave Thomas wrote:

On May 28, 2004, at 7:55, David A. Black wrote:


David A. Black
dblack@wobblini.net

“ts” decoux@moulon.inra.fr schrieb im Newsbeitrag
news:200405281300.i4SD0Le27873@moulon.inra.fr

Update: I’m now seeing a lot of news posts that are not making it to
the mail side. I’d thought it was more the other direction, but it
seems to be both.

It’s broken since 26/05/2004 – 16h00 (approx.)

UTC or which other time zone?
:wink:

robert

“Ben Giddings” bg-rubytalk@infofiend.com schrieb im Newsbeitrag
news:40B784F2.7050608@infofiend.com

Also, maybe the way the mailing list is set up can be changed. I
imagine right now that the way things work is:

Emails sent to the mailing list are sent out to all members,
one of those members is the gateway program, it sends them out
to the newsgroup. The newsgroup looks for new messages and when
it finds them, it emails them to the mailing list.

Maybe instead, we can have the mailing list be nothing but something
that copies messages to the newsgroup, and then have a secondary process
that takes everything from the newsgroup and sends it out to the mailing
list members. That way, instead of having two different lists that have
to be synchronized, we have only 1 (the newsgroup) that has to be
emailed out.

So you want to replace the direct mailing functionality by a detour via the
newsgroup. Isn’t that dangerous given that the news gateway seems to have
problems we can’t tackle? And, this might dissatisfy people because mails
won’t arrive as timely as they do right now. If all that could be dealt
with properly the idea doesn’t sound too bad to me. It would at least
eliminate the need for the gateway to check for message ids (or insert
something into message headers) to prevent endless looping. The question
remains whether the software at hand can be retrofitted to this behavior
with reasonable effort.

Kind regards

robert

Ben Giddings wrote:

For those of us who are interested in the matter, but aren’t aware of
how things work, can someone give a quick summary of how the mirroring
is supposed to work?

Some questions I have:

  • Is it a ‘fire and forget’ sort-of system – messages are sent to the
    NNTP server and if that doesn’t receive error messages it is assumed to
    have worked? Or does the software keep track of message IDs and somehow
    check to see that something has happened with them?

Your ‘fire and forget’ description makes me think you’ve been looking :slight_smile:

More specifically, it fires at nntp.rb and forgets. I think nntp.rb
would raise if there was a problem, though. The NNTP package has been
around for a while (4 years). It is (unmodified) from:
http://www02.so-net.ne.jp/~greentea/ruby/nntp/

  • How are messages mirrored from the newsgroup to the mailing list? Is
    it just something like “Add all new messages as of this date/time to the
    mailing list”, or does it keep track of message IDs in some way?

Add all new messages since the highest saved message-id.
Yes, it keeps track.
NG → ML direction seems to be flawless. [1]

  • Is there any ‘retry’ mechanism at all?

If you’re referring to “outside of NNTP protocol”, there seems to no
attempt to retry and if there is some reason for a message to be
missed, there’s no fancy recovery.

  • [DABlack] mentioned that if the message isn’t being copied by a
    not-directly-upstream server, then there’s nothing that can be done to
    trace it. It seems like there should be. If you can confirm that the
    gateway is working properly with the immediate NNTP server, find out
    what its neighbors are and ask the admins there to verify that the
    messages were transferred correctly, then if so, plead with them to
    check with their neighbors. I would suspect that if everything is
    properly mirrored in all those places, then it will get really hard to fix.

The easy way to annoy admins is to ask them to do work that we
should be doing. IME, they’re most helpful when there’s
evidence they can’t ignore :wink:
I’m confident that if the problem was further downstream than nntp.rb,
there would be losses from sources other than ours.

I don’t know much about NNTP, only the basics.

I’ll join you, there.

Is the way that our messages are added to the NNTP server the
same as the way that a standard network client does it? Are there extra
headers or anything that might make an NNTP server upset?

NNTP seems to be a sub-protocol of SMTP. I wondered if the
conversion from mail to news was leaving in headers that were,
strictly, SMTP only. But that doesn’t happen. Only selected headers
are copied, all are correct for NNTP.

BTW: You are asking all the right questions, IMO.

Also, maybe the way the mailing list is set up can be changed. I
imagine right now that the way things work is:

Emails sent to the mailing list are sent out to all members,
one of those members is the gateway program, it sends them out
to the newsgroup. The newsgroup looks for new messages and when
it finds them, it emails them to the mailing list.

Tick [or check in US].

One thing that bothers me is fixes to problems (e.g. floods) that
we’ve had in the past. There has to be a possibility that the
ruby-lang mailer has imposed restrictions on messages sent to the
“gateway ML member” and inadvertently introduced a corner-case.

Most of the serious people are on the ML, and if that’s where I
was viewing, I’d probably be wondering what the NG were all moaning
about. Mailing lists are reliable, Usenet isn’t? I think not.

This is a problem with the gateway although (as David pointed out
in a post that I just caught up on in the -talk archive because it
didn’t get through) that doesn’t mean the gateway software; rather,
it means the path from the ruby-lang mailer to the first Usenet
news server.

Maybe instead, we can have the mailing list be nothing but something
that copies messages to the newsgroup, and then have a secondary process
that takes everything from the newsgroup and sends it out to the mailing
list members. That way, instead of having two different lists that have
to be synchronized, we have only 1 (the newsgroup) that has to be
emailed out.

Any thoughts?

Yes, that’s interesting. The ML posters whose messages failed to get
through to c.l.ruby would have fairly immediate feedback that something
was wrong. Their messages wouldn’t appear.

Ben Giddings

I’m late in on this (one of my favourite topics), but my humble
investigation has been waiting for log evidence.

Logs are cool!
If rubygate[2] produced a log entry for any failed message, that
would provide a huge amount of information for me. I need to
know before I go chasing ghosts.

···

========================================================================

David Black said in [talk:101478]

My tests (fairly barebones, but anyway)
suggest that the gateway software is, in fact, passing along the
messages in question. Something is happening further downstream.

Software offers protection from accusations of inhumane treatment.
Pin the bug, then kill it.

  • Add regex to mail2news to exit unless the message is from
    dblack@address.net (insert your ML email address) else continue.

  • Start rubygate on your machine.

  • Reply to a message via ruby-talk

RESULTS (0=false; 1=true; *=don’t care):

  • [A] Appears in email inbox ?
  • [B] Logged_as_forwarded ?
  • [C] Appears on comp.lang.ruby ?

My humble ANALYSIS:

A - Assumed A1 because ML members see everything ?
I’m also assuming that your phantom posts appear
in your email inbox but not in c.l.ruby
Please say, if different.

B0:C0 Incorrect rubygate setup or mail2news oddity.

B0:C1 No, that didn’t happen :~>

B1:C0 suspect nntp.rb problem - OK, I’ll volunteer :frowning:
Please email the example input file for mail2news.

B1:C1 nntp.rb works - TU-Berlin news service oddity OR
“gateway ML member” (Dennis) is being censored
by ruby-lang mailer.

Modifications/criticism welcome. I want to see it fixed, too.

daz

[1] As long as the news service continues incrementing
and doesn’t reset to a lower number (ask PragDave;)

[2] (dunno if it was Dennis or David or the “Washington Post” newspaper
who came up with that fine name for the gateway software.)

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

···


David A. Black
dblack@wobblini.net

UTC or which other time zone?

MTZ : moulon time zone

Guy Decoux

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

What is the difference : the header Sender: ?

Guy Decoux

Robert Klemme wrote:

Add all new messages since the highest saved message-id.
Yes, it keeps track.
NG → ML direction seems to be flawless. [1]

This doesn’t seem to fit Martin’s analysis: there seem to be messages
missing on both sides, don’t they?

There were problems last year which were fixed by using the more
reliable feed. I think there was a brief problem in the NG → ML
direction very recently. If folks are trying to fix things, this
can be expected. Martin’s data are too broad. The problem we’re
addressing started around Feb. this year.

Can we really be sure that message ID’s follow this pattern?

The message-id used is numeric; local to the news service. It’s nothing
to do with those strings in the mail headers.

I’m confident that if the problem was further downstream than nntp.rb,
there would be losses from sources other than ours.

Do we know that there aren’t?

The hardcore Usenet crew are really vocal when things go wrong with their
communication network. I don’t know that there aren’t problems but I
can’t contemplate that a serious snag with propagation would go unreported
for 4 months.

NNTP seems to be a sub-protocol of SMTP.

As I understand NNTP […] rather different in nature.

Maybe copying is not sufficient, maybe some header contents have to be
massaged in order to fit with NNTP.

<from RFC 1036>
“The USENET News standard is more restrictive than the Internet standard,
placing additional requirements on each message and forbidding use of
certain Internet features. However, it should always be possible to use
a tool expecting an Internet message to process a news message.”
</>

One thing that bothers me is fixes to problems (e.g. floods) that
we’ve had in the past. There has to be a possibility that the
ruby-lang mailer has imposed restrictions on messages sent to the
“gateway ML member” and inadvertently introduced a corner-case.

But wouldn’t that mean loss of postings in the mailing list as well? I
mean, if the mailer did selective forwarding of messages then everybody (NG
readers as well as ML readers) would miss them.

No?
Send to all ML members except the g/way, because the g/way is broken
and it might mirror them back to the ML resulting in the ML getting the
messages twice.
Or this … http://www.ruby-talk.org/92529 ?
(I have no idea what happened here.)

This is a problem with the gateway although (as David pointed out
in a post that I just caught up on in the -talk archive because it
didn’t get through) that doesn’t mean the gateway software; rather,
it means the path from the ruby-lang mailer to the first Usenet
news server.

Do you mean the selective transmission works only for mail → news and not
mail → mail? Now I’m confused here…

Not by me, surely :wink:
All ML members post to a single recipient ruby-talk@ruby-lang.org. It’s
up to the bulk mailer where it goes next. The g/way is just one member.
One scenario is that the missing messages are not getting to the gateway.
If they did, they would be forwarded to c.l.ruby like the rest.
That’s my favourite suspicion and could easily be disproved by a single
log entry for a message that failed to make the NG.

With this single piece of evidence (which shouldn’t be so difficult to
obtain …) we have instant progress rather than looking at every other
non-possibility (which I, and others, have already done).
If the log says a phantom message has been forwarded, it cuts the
upstream right out of the equation and IMHO leaves nntp.rb only.
Debugging is a lot easier when you know you’re focusing on the right area
and it’s a pointless waste of time when you strongly suspect that the
problem lies elsewhere.

Kind regards

robert

To you too,

daz

David A. Black wrote:

[…] testing again… small meaningless tweak to headers…

Mirrored using mail2news.rb (the version you sent me - unmodified).

Worked as I expected, so that rules out nntp.rb & headers.

Took your message out of my inbox (from ML).
Windows box, so newlines are different.
I might gsub them and send again.

Cheers,

daz

Hi –

···

On Tue, 1 Jun 2004, ts wrote:

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

What is the difference : the header Sender: ?

Yes. I’m not claiming it’s an intelligent idea :slight_smile: Just trying
everything, since I have no idea what the problem is.

David


David A. Black
dblack@wobblini.net

Hi –

···

On Tue, 1 Jun 2004, daz wrote:

All ML members post to a single recipient ruby-talk@ruby-lang.org. It’s
up to the bulk mailer where it goes next. The g/way is just one member.
One scenario is that the missing messages are not getting to the gateway.
If they did, they would be forwarded to c.l.ruby like the rest.
That’s my favourite suspicion and could easily be disproved by a single
log entry for a message that failed to make the NG.

I’ve got a request in to Dennis for log files, but I am almost certain
he had indicated to me at some point that that wasn’t happening.
Anyway hopefully we’ll get concrete confirmation.

David


David A. Black
dblack@wobblini.net