[ntp:questions] ntpd, boot time, and hot plugging

Per Hedeland per at hedeland.org
Sun Feb 6 21:59:52 UTC 2005


In article <mailman.59.1107716992.583.questions at lists.ntp.isc.org> Brad
Knowles <brad at stop.mail-abuse.org> writes:
>At 12:01 PM +0000 2005-02-06, Per Hedeland wrote:
>
>>  Sorry, even though I could, I don't see why I should spend time and
>>  effort debugging the software that you are using to inject malformed
>>  articles into the group.
>
>	If you're going to make claims of broken threading, then you need 
>to provide evidence of that

OK, here's an example - it's actually you following up to yourself:-)
(in another thread):

Message-ID: <mailman.46.1107539389.583.questions at lists.ntp.isc.org>
References: <1107532000.861193.51530 at f14g2000cwb.googlegroups.com>
        <p0620076bbe295454e067@[10.0.1.2]>

Now, there has been no message with a Message-ID of
<p0620076bbe295454e067@[10.0.1.2]> posted to the group (or any other
group) - the message you are responding to had Message-ID
<mailman.43.1107536028.583.questions at lists.ntp.isc.org> when posted.
I.e. the References header is "wrong" (trn manages to connect the
message to the thread anyway thanks to the first Message-ID in
References, but obviously can't connect it at the right place, and
indicates that the <p0620076bbe295454e067@[10.0.1.2]> message is
missing).

I'm writing "wrong" intentionally, because closer examination reveals
that this isn't the real problem (in this case at least) - it turns out
that *all* messages coming from the mailing list have Message-IDs of the
<mailman...questions at lists.ntp.isc.org> form. I.e. something (Mailman?)
is actually *changing the Message-ID* of the posted messages, which is a
really horrible thing to do, and which has consequences way beyond
broken threading in a newsgroup.

Whether it does this just for the messages gatewayed to the newsgroup or
also on the mailing list side I don't know, but other evidence suggests
that it's specific to the newsgroup gatewaying - e.g. there are
followups from other people on the mailing list that have the type of
Message-ID your MUA produces as the immediate parent in their References
headers.

>	Perhaps it is trn that is broken in this case?  After all, it is 
>pretty ancient software, and the algorithm for detecting threads has 
>most likely been updated a few times since the last time it was 
>touched.

Sorry, but there hasn't been any advances in that technology - mostly
because trn more than a decade ago pushed it to the very edge of what is
possible. Recent newsreaders are generally inferior in that respect
(unless you consider unconditionally hiding the brokenness of the
threads as a feature).

--Per Hedeland
per at hedeland.org



More information about the questions mailing list