Pan Date header preference setting for time zone

On Sun, 03 Jul 2016 11:33:24 -0700, Mike Easter wrote:

In the very old news from 2000, Pan addresses a number of Date: related
bugs which were fixed in those old .8 series versions.

I saw that.
I never said I knew what I was talking about though!

I think I am beginning to know what is happening.

http://pan.rebelbase.com/oldnews/oldnews-08.html (see search page on
Date Pan says it inserts Date.

Yep. I believe you.
I do not think the user has *any* control over the date.

However ... if Pan is inserting the date, then why did Mixmin have a
negative -0000 time when aioe and netfront had a positive +0000 time?

Perhaps it's because Mixmin uses port 563/SSL and Aioe and Netfront used
port 119?

However, that doesn't explain how we got a "+0000 (UTC)" using telnet
only, without inserting the date!

telnet nntp.aioe.org 119
help
post
- 340 Ok, recommended message-ID
References:
Message-ID:
From: Henry Jones
Newsgroups: news.software.readers
Subject: Pan Date header preference setting for time zone

This is a test of not specifying the date in telnet to aioe.
..
quit

0.140 is a different 'pan2 era'. Old v./s
http://pan.rebelbase.com/download/releases/

Huh? I'm confused.
I downloaded the very *latest* Pan (or so I thought).
The latest, as of 2016, is version 0.140.
Is it not?

Anyway, it's pretty clear what is happening, although I can't explain it.
The simplest solution is to telnet into aioe and specify the date.

Three tests tell us what we need to know.
1) Telnet and do not specify a date or comment.
2) Telnet and specify a +0000 date & comment.
3) Telnet and specify a -0000 time & comment.

1) When I do not specify a date, aioe inserts the + and the (UTC)
+0000 (UTC)
2) When I specify the + date, I get exactly what I specify
+0000 (comment)
3) When I specify the - date, I get exactly what I specify
-0000 (comment)

Since Pan is not involved when I telnet to nntp.aioe.org:119, either my
operating system (WinXP) is giving aioe the date, or, more likely, aioe
*knows* the date, and it's aioe who inserts the "+0000 (UTC)".

Are any of you using aioe?
Can you test this out?

Henry Jones wrote:
I think I am beginning to know what is happening.

I think that Pan is inserting the Date and I think the best/ most
consistent test/challenge would be to post a properly configured Date.

I believe that if a Date line is not included, there will be more
variability in what different news servers do.

Given that a Date is going to be inserted, the experimenter would choose
how to vary the Date, such as + or - 0000 or even one with a tz offset.

--
Mike Easter

On Sun, 03 Jul 2016 13:22:17 -0700, Mike Easter wrote:

I think that Pan is inserting the Date and I think the best/ most
consistent test/challenge would be to post a properly configured Date.

I believe that if a Date line is not included, there will be more
variability in what different news servers do.

Given that a Date is going to be inserted, the experimenter would choose
how to vary the Date, such as + or - 0000 or even one with a tz offset.

I still do not understand how to interpret my experiments, but I don't
think it's Pan, per se, which is inserting the date.

It *might* be Pan inserting the date; but the date *always* gets inserted
whether or not Pan puts it there (which was proven in the telnet
experiments).

I can think of only 4 ways the date can get the

0. The user manually inserts the date (which seems impossible with Pan).
2. Pan inserts the date (which flies against the evidence of mixmin having
a different date format than aioe/netfront using the same Pan application!)
3. The OS is inserting the date (in my case, this being WinXP) which might
fit the evidence because mixmin uses a different OS commands, I think,
than does aioe/netfront.
4. The server inserts the date (which we *know* to be true because when I
telnet and don't provide a date, a date is provided for me).

The evidence is clear.
But the interpretation of the evidence is not.
So I may be wrong and I await someone who knows more than I do to make
sense of why the telnet sessions turned out the way they did.

If anyone knows how to telnet to mixmin 563, please tell me how to handle
the certificate. I suspect if we can telnet to mixmin, we will solve the
enigma instantly.

On +0000, Whiskers wrote:

Very unlikely. No news-server should touch a header provided by a user;
pass it on unchanged, drop it completely, or reject the post, but never
ever change anything.

I don't claim to understand this stuff but I don't think *anyone* here
does either.

These are facts:

1. I telnet to aioe from WinXP with a US Timezone and if I don't put a
date, a time & comment of the following format is put there for me by
*something*.
Sun, 03 Jul 2016 14:20:42

2. I telnet to aioe from WinXP and whatever I put as a date shows up
verbatim, whether I put:
Sun, 03 Jul 2016 14:20:42 +0000 (comment)
Sun, 03 Jul 2016 14:20:42 +1234
Sun, 03 Jul 2016 14:20:42
etc.

3. In Pan, there is no way I know of to set the date, but if I use aioe or
netfront, I get a date format of:
Sun, 03 Jul 2016 14:20:42 +0000 (UTC)

4. Yet, in Pan, if I go to Mixmin (port 563), I get a date format of:
Sun, 03 Jul 2016 14:20:42 -0000 (UTC)

Those are all easily verified facts.
If anyone can explain them, I'm all ears.

On Sun, 03 Jul 2016 11:47:15 -0700, Mike Easter wrote:

During the EL in-house nntp days, it ALWAYS replaced the user's Date
header content with its own accurate time/date values, thus avoiding the
possible 'problem' of a user's Date being wrong, which must've caused
some kind of trouble at EL's end. I guess.

When I was telnetting into aioe, I was experimenting with various DATE
formats, and half of them failed because of either syntax errors or they
were too far off from aioe's own internal clocks.

I must have gotten at least 3 or 4 *different* messages when the syntax
was correct, but the dates I entered in were too far off so there is a
*lot* of error checking going on at the server level.

So it is writ, so mote it be....
Henry Jones :
On +0000, Whiskers wrote:

Very unlikely. No news-server should touch a header provided by a user;
pass it on unchanged, drop it completely, or reject the post, but never
ever change anything.

I don't claim to understand this stuff but I don't think *anyone* here
does either.

These are facts:

1. I telnet to aioe from WinXP with a US Timezone and if I don't put a
date, a time & comment of the following format is put there for me by
*something*.
Sun, 03 Jul 2016 14:20:42

2. I telnet to aioe from WinXP and whatever I put as a date shows up
verbatim, whether I put:
Sun, 03 Jul 2016 14:20:42 +0000 (comment)
Sun, 03 Jul 2016 14:20:42 +1234
Sun, 03 Jul 2016 14:20:42
etc.

3. In Pan, there is no way I know of to set the date, but if I use aioe or
netfront, I get a date format of:
Sun, 03 Jul 2016 14:20:42 +0000 (UTC)

4. Yet, in Pan, if I go to Mixmin (port 563), I get a date format of:
Sun, 03 Jul 2016 14:20:42 -0000 (UTC)

Those are all easily verified facts.
If anyone can explain them, I'm all ears.

When you don't supply a Date: header the server will supply one.
I think this has been explained a few times in this thread.

When the server supplies the header in UTC it will obey the rules
regarding '+' or '-' as they apply to the server supplying the
header, not the newsreader that isn't supplying the header.

Those rules have been discussed in this thread as well.

What hasn't been spelled out in a simple way is that the server
builds the Date: header using those rules based on its own clock
and time zone settings, not on the one not supplied by the news
reader.

Could it be that Pan doesn't supply the Date: header at all? I
use slrn with leafnode2 as the local news server and that header
seems to be coming from leafnode2.

Mike "step back and think it out" Yetto
--
"The important thing in science is not so much to obtain new facts as to
discover new ways of thinking about them."
- Sir William Henry Bragg, Nobel Prize for Physics, 1915

On 2016-07-03, Henry Jones wrote:
On Sun, 03 Jul 2016 13:22:17 -0700, Mike Easter wrote:

I think that Pan is inserting the Date and I think the best/ most
consistent test/challenge would be to post a properly configured Date.

I believe that if a Date line is not included, there will be more
variability in what different news servers do.

Given that a Date is going to be inserted, the experimenter would choose
how to vary the Date, such as + or - 0000 or even one with a tz offset.

I still do not understand how to interpret my experiments, but I don't
think it's Pan, per se, which is inserting the date.

It *might* be Pan inserting the date; but the date *always* gets inserted
whether or not Pan puts it there (which was proven in the telnet
experiments).

I can think of only 4 ways the date can get the

0. The user manually inserts the date (which seems impossible with Pan).
2. Pan inserts the date (which flies against the evidence of mixmin having
a different date format than aioe/netfront using the same Pan application!)
3. The OS is inserting the date (in my case, this being WinXP) which might
fit the evidence because mixmin uses a different OS commands, I think,
than does aioe/netfront.
4. The server inserts the date (which we *know* to be true because when I
telnet and don't provide a date, a date is provided for me).

The evidence is clear.
But the interpretation of the evidence is not.
So I may be wrong and I await someone who knows more than I do to make
sense of why the telnet sessions turned out the way they did.

If anyone knows how to telnet to mixmin 563, please tell me how to handle
the certificate. I suspect if we can telnet to mixmin, we will solve the
enigma instantly.

Certainly if the user doesn't supply a Date header the server will, just
as happens with the Message-ID header. If the server doesn't think the
user-supplied Date header is 'correct' then it should either reject the
post or insert a separate header of its own, 'NNTP-Posting-Date'. Some
servers always insert that header.

When I had a dial-up internet connection, I used to post via a local
caching proxy news-server (Hamster or Leafnode), dialing up to log in to
the news-server at regular intervals or when I'd got a batch of posts to
send. That could result in my posts appearing with a wide range of Date
headers (several hours sometimes) but with NNTP-Posting-Date headers all
at the same time or within a minute or so. Keen eyed geeky readers
sometimes commented on this. I believe my local proxy server (running
on my own computer) was inserting the Date headers when I wrote the
articles; the NNTP-Posting-Date was being supplied by the upstream
server I was posting to.

--
-- ^^^^^^^^^^
-- Whiskers
-- ~~~~~~~~~~

Whiskers wrote:
Keen eyed geeky readers

Volunteer clock police courtesy notice :-)

--
Mike Easter

on 7/3/2016, Whiskers supposed :
On 2016-07-03, Henry Jones wrote:
On Sun, 03 Jul 2016 13:22:17 -0700, Mike Easter wrote:

I think that Pan is inserting the Date and I think the best/ most
consistent test/challenge would be to post a properly configured Date.

I believe that if a Date line is not included, there will be more
variability in what different news servers do.

Given that a Date is going to be inserted, the experimenter would choose
how to vary the Date, such as + or - 0000 or even one with a tz offset.

I still do not understand how to interpret my experiments, but I don't
think it's Pan, per se, which is inserting the date.

It *might* be Pan inserting the date; but the date *always* gets inserted
whether or not Pan puts it there (which was proven in the telnet
experiments).

I can think of only 4 ways the date can get the

0. The user manually inserts the date (which seems impossible with Pan).
2. Pan inserts the date (which flies against the evidence of mixmin having
a different date format than aioe/netfront using the same Pan application!)
3. The OS is inserting the date (in my case, this being WinXP) which might
fit the evidence because mixmin uses a different OS commands, I think,
than does aioe/netfront.
4. The server inserts the date (which we *know* to be true because when I
telnet and don't provide a date, a date is provided for me).

The evidence is clear.
But the interpretation of the evidence is not.
So I may be wrong and I await someone who knows more than I do to make
sense of why the telnet sessions turned out the way they did.

If anyone knows how to telnet to mixmin 563, please tell me how to handle
the certificate. I suspect if we can telnet to mixmin, we will solve the
enigma instantly.

Certainly if the user doesn't supply a Date header the server will, just
as happens with the Message-ID header. If the server doesn't think the
user-supplied Date header is 'correct' then it should either reject the
post or insert a separate header of its own, 'NNTP-Posting-Date'. Some
servers always insert that header.

In the versions of Pan that I have used, you could suppress the Date
header and the server would put its own there. I never considered it a
privacy issue to do it either way.

Henry Jones explained on 7/3/2016 :
On Sun, 03 Jul 2016 11:47:15 -0700, Mike Easter wrote:

During the EL in-house nntp days, it ALWAYS replaced the user's Date
header content with its own accurate time/date values, thus avoiding the
possible 'problem' of a user's Date being wrong, which must've caused
some kind of trouble at EL's end. I guess.

When I was telnetting into aioe, I was experimenting with various DATE
formats, and half of them failed because of either syntax errors or they
were too far off from aioe's own internal clocks.

I must have gotten at least 3 or 4 *different* messages when the syntax
was correct, but the dates I entered in were too far off so there is a
*lot* of error checking going on at the server level.

Message IDs can be supplanted also if the user supplied ones don't pass
muster.

On 2016-07-03 17:24, Henry Jones wrote:
I could test better if I knew how to *telnet* to one of these no-
registration free newsservers, so that I would not be using a news server
at all, that would help us determine whether the news server or the news
agent is adding that ± sign.

Better use a server on your local network. in Linux the easiest to setup
is probably leafnode.

--
Cheers,
Carlos E.R.

On Sun, 03 Jul 2016 19:38:46 -0400, FromTheRafters wrote:

Message IDs can be supplanted also if the user supplied ones don't pass
muster.

Being a novice at telneting into newssservers, I first crafted my own
Message IDs, but I ran into too many syntax-checking failures on aioe.

But using the suggested Message-ID worked every time.

The *problem* I had was how to *reply* to a given message!
I didn't see any "reply" command in the nntp "help"!

If I posted with a "Re", the post was rejected with a clear error (there's
a *lot* of error checking going on, even with telneting into the server!).

So I had to add a *"References"* header.

But how the heck do you build them?
I couldn't figure them out.

I tried copying message IDs, but why on earth are there more than one
References Message-IDs in each reply? (How can one be replying to more
than one message at a time anyway?)

How does one properly construct the "References" line in telnet?

On 2016-07-04 01:36, FromTheRafters wrote:

In the versions of Pan that I have used, you could suppress the Date
header and the server would put its own there. I never considered it a
privacy issue to do it either way.

The OP feels it is a privacy issue because people can guess that he is
using Pan by looking at the Date header. The contention is that only Pan
writes "(UTC)" in the header, and thus, when people see that style of
date writing they will guess "he is using Pan!". And giving out that
tiny bit of information "I use Pan" is that privacy issue.

But either Pan or more probably, his nntp server, inserts a
"NNTP-Posting-Host" header, and this one gives away his exact position
at the time of posting!

Even if that were not the case, he is using a real mail address,
apparently, and besides that, surely by looking at all the headers it
should be possible to guestimate the software chain used to create and
post that piece. And by analyzing the text body content, the person that
wrote it can also be guessed, even if the mail address is faked.


Resistance is futile. You will be assimilated.


--
Cheers,
Carlos E.R.

On 2016-07-03 23:47, Henry Jones wrote:
When I was telnetting into aioe, I was experimenting with various DATE
formats, and half of them failed because of either syntax errors or they
were too far off from aioe's own internal clocks.

I must have gotten at least 3 or 4 *different* messages when the syntax
was correct, but the dates I entered in were too far off so there is a
*lot* of error checking going on at the server level.

Some servers refuse to accept a post if the date is too old.

--
Cheers,
Carlos E.R.

Henry Jones presented the following explanation :
On Sun, 03 Jul 2016 19:38:46 -0400, FromTheRafters wrote:

Message IDs can be supplanted also if the user supplied ones don't pass
muster.

Being a novice at telneting into newssservers, I first crafted my own
Message IDs, but I ran into too many syntax-checking failures on aioe.

But using the suggested Message-ID worked every time.

The *problem* I had was how to *reply* to a given message!
I didn't see any "reply" command in the nntp "help"!

If I posted with a "Re", the post was rejected with a clear error (there's
a *lot* of error checking going on, even with telneting into the server!).

So I had to add a *"References"* header.

But how the heck do you build them?
I couldn't figure them out.

I tried copying message IDs, but why on earth are there more than one
References Message-IDs in each reply? (How can one be replying to more
than one message at a time anyway?)

How does one properly construct the "References" line in telnet?

It has to do with threading. It's just a list or chain of Message-IDs
of posts going back to the OP. I suppose you can use the Message-ID for
that too.