[Mailman-Users] Structure of Mailman archives

[Mark Sapiro]

Diana Kirk wrote:

>I've gone over the FAQ on moving archives. Since I don't have actual 
>access to /archives on my system (shared hosting), I'm going to have 
>to ask Support to do what's necessary. I can make sure everything is 
>in mbox format. However, I have archives going back to 1996 from 
>several different mailing list programs (listserv, majordomo, and 
>mailman).
>
>Specifically, my question is:  How is 
>/archives/private/[list].mbox/[list].mbox actually structured? The 
>FAQ makes it sound like it is one giant mbox file.


Answered in another reply, but Yes.


>Also, there are two 
>[list].mbox in the path. The first is a subdirectory of 
>/archives/private. The second is the actual mbox file. Are they both 
>really the exact same name?


Also answered, Yes.


>I have new volume numbers for each year, and the archives are set to 
>show all posts per month. Does /archives/private reflect that via 
>subdirectories? For example, 
>/archives/private/[list]/2005-October/[message_number].html would be 
>the path for a current message. Is that how I should have the past 
>posts uploaded? That is, 
>/archives/private/[list]/year-month/[list].mbox where I've zipped the 
>file so that the year-month is a subdirectory containing an mbox with 
>the messages for that time period. 


This is how the archives are structured, but all this structure is in
addition to the cumulative mailbox file
/archives/private/[list].mbox/[list].mbox. All you need to (re)build
an entire Pipermail arvhive is the one cumulative mbox file.


>Or should I give Support one big mbox with all the messages, the 
>instructions on what to do from the FAQ, and assume Mailman will 
>figure it out?


If you're trying to create one huge archive containing all the various
listserv, majordomo and mailman archives, the thing to do is to make
the one huge cumulative mbox file. As long as this has proper "^From "
lines with good dates and good Date: headers, "bin/arch --wipe" will
build the Pipermail archive structure from it. The archives will also
be thereaded according to Message-Id:, In-Reply-To: and References:
headers in the mbox file.


>Am I out of my mind thinking this can be done? I'll take a "no 
>comment" as a "yes" :). I would really like to do this if it's 
>possible.


The only problem is insuring the integrity of the mbox file. There is a
bin/cleanarch script that attempts to insure that messages in the mbox
file do not themselves contain "^From " lines. This is not perfect,
but it can help.

-- 
Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan