[svlug] Missing Web archive to be added: 1997 to 2000

Rick Moen rick at linuxmafia.com
Thu Sep 22 21:40:12 PDT 2011


Quoting Don Marti (dmarti at zgp.org):

> Let me guess -- the archive restoration process
> somehow involved tooling up and down Hwy. 101 in a
> silver Honda?

The lists.svlug.org mailing list host has had an mbox file left over
from the Majordomo setup sitting around, all along, and nobody bothered
to mention it or do anything with it.  I came across it by accident.

The obstances to merging it in were:

1.  Majordomo did an even worse job than Mailman of correcting 
unescaped body-text lines beginning with 'From '.  I figured out 
how to script a repair using sed.

2.  Many posts in the Majordomo mbox had Content-Type: headers that gave
the Mailman Web archiver script indigestion.  Notably, the ones with
just 'Content-Type: text' were misparsed as having binary attachments.
Again, a bit of work with sed fixed that.

3.  A number of subscribers' posts with severely inaccurate system
clocks got archived in wrong or ludicrously wrong months, such as
January 1980.  I manually edited the Date: headers on a half dozen or so
that stuck out.  (Mailman can be configured to override _upon initial
receipt_ Date: headers that are severely wrong, but that wouldn't have
helped in this scenario, anyway.

4.  Several subscribers had been in the habit of posting forwards from
elsewhere with all original headers intact, and (again) Majordomo didn't
bother to escape the body-internal 'From ' SMTP envelope line, which
wreacked holy hell with Mailman's archiver.  A bit of detective work
found that.


Anyhow, I ran the real regeneration of the production
svlug at lists.svlug.org mailing list archive tonight, and it's all fixed
and complete, now.

It may be of interest that SVLUGgers have posted 57,311 messages to this
mailing list since its founding.  (The sheer volume of back postings
plus the above-cited obstacles are what made this a bit of a
production.)




More information about the svlug mailing list