Stephen Gran wrote:
> So I've looked through a few weeks of mail logs to packages.debian.org,
> and it looks like it collects some useful mail from automated scripts
> on various debian.org machines (primarily ries), and about 1000 spams a
> day from elsewhere. I haven't done an exhaustive survey, but it seems
> pretty clear so far that the domain does not get any significant amount
> of legitimate mail from machines other than the debian.org hosts.
It's not uncommon for package's to have documentation that says to mail
to p.d.o. See for example aptitude, base-passwd, debian-faq,
developers-reference.
Some packages even emit it at runtime:
cron@packages.debian.org fscked up. Send him a nasty note [1]
How odd.. The sizes didn't match, email apt@packages.debian.org
print_fill('Printer on %s was detected by Debian using the ad-hoc method. Please submit the following information to foomatic-db@packages.debian.org:', device)
Other packages, including reportbug --kudos, BUGBASH, and debconf-updatepo
directly send mail there.
Personally, I generally use it as per its original intended use; for contacting
a package's maintainer w/o looking up their actual email address; generally
when CCing them about a related bug report. More often, someone will CC a bug
report to me via p.d.o, or contact me directly via those addresses.
This gives a pretty good idea of my volume, though it won't list all mails.
No mails listed here are from automated scripts AFAIK.
joey@gnu:~/mail/archive>echo "year\tin\tout"; for y in $(seq 1997 2009); do echo "$y\t$(zegrep '^To:.*@packages.debian.org' inbox/$y-* |wc -l)\t$(zegrep '^To:.*@packages.debian.org' Sent/$y-* |wc -l)"; done
year in out
1997 14 4
1998 53 20
1999 37 26
2000 24 16
2001 51 20
2002 36 16
2003 17 14
2004 12 3
2005 15 1
2006 151 4
2007 151 6
2008 134 1
2009 35 0
And here's the per-package breakdown:
joey@gnu:~/mail/archive>zegrep '^To:.*@packages.debian.org' inbox/* Sent/* | perl -ne 'while (/([-\w]+)\@packages.debian.org/g) { print "$1\n" }' |sort|uniq -c | sort -rn| head -n 20
91 debhelper
67 debconf
43 ikiwiki
37 alien
24 devscripts
21 debian-maintainers
16 etckeeper
15 nslu2-utils
14 sleepd
14 mr
13 lintian
12 pdmenu
11 archivemail
10 znc
10 xaw-wrappers
10 pristine-tar
10 jetring
10 dpkg-repack
# 195 more packages not listed
Anyway, this feels like it may be a false optimisation to me. You know
that p,d.o gets 1000 spams a day, but you don't know how many spams a
day are sent directly to each of the email addresses that p.d.o forwards
to. My suspicion is that each *individual* package maintainer email
address could easily be getting 1000 spams a day. If on average one or
two more spam come through p.d.o, that's very minor. Unless spam coming
through p.d.o is somehow harder to filter?
--
see shy jo
[1] Disappointingly, based on strings, this specific example is optimised
out of the actual crontab binary.
Attachment:
signature.asc
Description: Digital signature