Thanks for having a look around! I think your findings make a nice case for retaining the original archive.
>>>>> On Wed, 12 Jul 2023 23:55:05 +0200, Erik Huelsmann said:
>
> The web interface for the new mailman version is at
> https://mailman3.common-lisp.net/postorius/ for mailing list administration
> and https://mailman3.common-lisp.net/hyperkitty/ for the archives.
>
> Let me know if you find irregularities or have other remarks.
It looks nice.
I had a play with the new archives and found something strange with this
thread:
https://mailman3.common-lisp.net/hyperkitty/list/tbnl-devel@common-lisp.net/thread/AB3VMJNMVYAIKVEDJQNERCRRUFO4JGYN/
There are two problems:
1. It appears to have only has two messages, one from Jingtao Xu and a reply
from Hans Hübner, However, there is another thread in the archive that is a
reply to Hans Hübner's message:
https://mailman3.common-lisp.net/hyperkitty/list/tbnl-devel@common-lisp.net/thread/OT46NCYD5CHDZPPQVHP6TEFJ2BGGNNQE/
Downloading the threads shows matching message-id and in-reply-to headers
(<CABj0bw+=GVc60=Y7F1OX=9F6us5KN6cZvd1yx6xsURsaQa8sKA@mail.gmail.com>) so why
is it a separate thread?
I have no idea at all. In the original archive, they're indeed all part of a single thread. I had no trouble using the administrative interface to merge the various threads into a single one, but it was manual work indeed. I did search around on the web, but didn't find any information or complaints about others running into the same problems...
2. The dates displayed for the messages are in April 2015, but the messages
themselves quote dates from May 2013.
The old pipermail archives (from May 2013) don't show these problems:
https://mailman.common-lisp.net/pipermail/tbnl-devel/2013-May/thread.html#3703
This one I understand better: the migration procedure suggests to check the historic mboxes for import problems, including date formatting issues. The pre-import check script I was supposed to use did change(fix?) some dates; apparently, not all for the better. There's two ways forward on this one: we can leave them as-is and keep hosting the original archives. Or I could throw away the original imports and re-import (hoping the import scripts didn't fix more than incorrectly fixing these dates). I'm inclined to do the former, because I know that there were a few mails discarded for not being ascii-only mail headers (possibly someone used UTF8 or Latin1 characters in their name, which isn't allowed by the RFCs).
Regards,
Erik.