Mailing List Server Date
Sat, 15 Jan 2000 11:47:29 +0000 (GMT)
On Fri, 14 Jan 2000, Brian K. Shiratsuki wrote:
> Hypermail now produces correct dates for the January 2000 ns-users archives.
I see mash's copy of hypermail has been upgraded to 2b28, and that's
been run on January 2000. Thanks, Brian.
The December 1999 archive index created with hypermail 2b3 is
particularly incomplete, and certainly needs regenerating to list more
than the 158 messages it lists. There are 237 messages rendered to
HTML pages in the month, but the index stops at 8 December, even
though the last line on the page says the archive was generated on 31
Can you regenerate the December 1999 archive from an archive mailbox
using same, please? That should bring 'missing' messages in this month
to light. (reading warnings on http://www.hypermail.org/y2k.html
here, which might account for the 8/31 Dec discrepancy above.)
I did some spot counts on other archive pages, and the results are
worrying; the hypermail 1.02 index generator appears to be lossy
too. For example:
November 1999 (the last with hypermail 1.02) has 456
messages archived (0000.html - 0455.html). There are 458 lines of
links on the index page according to grep -c href (each message title
on a separate line), so if you subtract six lines to account for the
context non-message links at top and bottom (more than one link a
line), four messages remain unlisted.
October 1999 has 419 messages stored as individual webpages
(0000.html - 0418.html), while grep -c \<li\> 9910index.html
gives 418 - but subtract the five non-message bullet points at top
and bottom, and you're down to 413; five messages apparently not
listed in the index. One of them is 0158.html. Another is 0342.html.
The missing messages I located are both followups; I suspect that
hypermail 1.02's threading code is unreliable, and that full
regeneration of the archive would be prudent.
(I suppose that, in grand IETF tradition, I should also ask for
gzipped mailboxes to be made available from the ftp server so that
the truly paranoid can generate their own indices.)
is hoping that full regeneration fixes the two-digit years on: