At 04:33 29/01/2004, Matt Kettler wrote:
At 02:00 AM 1/28/2004, Simon Byrnand wrote:
Has anyone else noticed frequent timeouts with Razor2 ?
I disabled it Friday due to timeouts.
Ah... Yes I just disabled it myself last night after about half an hour of
debugging to make sure it wasn't a sof
uldn't see anything about it over on the razor
homepage or mailing list archives
Regards,
Simon
---
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of E
ip address of
the server immediately prior to it.
Once you realise this, you can see how whitelisting is easy to fool...(and
why spammers have a field day including bogus headers...)
Regards,
Simon
---
This SF.net email is sponsored by: IBM
Matt,
On Mon, 29 Dec 2003, Matt Kettler wrote:
> At 02:50 PM 12/29/2003, Simon Matthews wrote:
> >Your comment made me look into the issue a little more and I see that a
> >mail server is listed as "trusted" when I don't think it should be. In the
> >email
Matt,
At 02:28 PM 12/29/03 -0500, Matt Kettler wrote:
At 12:17 PM 12/28/2003, Simon Matthews wrote:
Specifically, the RCVD_IN_DYNABLOCK
check. Note that 192.168.10.250 is a local (within the LAN) relay.
If you're going to use 192.168.*.* networks, add them to your
trusted_networks statemen
Matt,
Thanks for the suggestion.
I checked in the logfiles and it looks like the 192.168.10 domain is
already treated as trusted (ie. spamassassin infers automatically that it
is trusted).
I see lines in the logfile such as:
debug: received-header: relay 192.168.10.250 trusted? yes
Simon
At
At 15:11 28/12/2003 -0500, David A. Roth wrote:
On Sunday, December 28, 2003, at 01:40 PM, Simon Byrnand wrote:
I upgraded from 2.60 to 2.61 and I am getting many false positives. It
seems that Bayes is pushing it with a score of 5.4. What are people to
do to get around this? Do you set Bayes
attitude a lot.
SpamAssassin is not a commercial company which you have paid money to for
support, it is the efforts of a large group of individuals, most of them
working on their free time.
Of all the open source software that I use, SpamAssassin has by far the
best mailing list and most respons
another false positive with 4.9.
Err, BAYES_00 is *minus* 4.9 :) And is correct if the message is ham. If
it still got flagged as spam, it must have had many other tests that
fired...without seeing the headers, we can only speculate...
Regards,
Simon
d from network); 20 Dec 2003 00:02:34 -
Received: from unknown (HELO Arnold) (:[EMAIL PROTECTED])
by 205-158-62-78.outblaze.com with SMTP; 20 Dec 2003 00:02:34 -
Message-ID: <[EMAIL PROTECTED]>
Reply-To: "Arnold Matthews"
From: "Arnold Matthews"
To: "Ed Mat
send HTML in this format
The IP address 192.168.10.250 refers to a local relay. It should not
however have shown the email as directly received from 81.152.14.12,
which is the check "RCVD_IN_DYNABLOCK".
Thanks!
Simon
-- Forwarded message --
Return-Path:
X-Original-
tely, but that means calling
spamc twice, something I can't really afford to do when our server is under
seige already.
Anyone else have any thoughts ?
Regards,
Simon
---
This SF.net email is sponsored by: IBM Linux Tutorials.
Become a
it without consequences they will.
If there is a strong risk of being caught and hauled off to jail, many
others will think twice.
I think that day is not far away now...
Regards,
Simon
---
This SF.net email is sponsored by: IBM Linux Tutorials.
reduce disk activity, which would be just as useful) by
somehow getting spamd to create these files on a specified ram filesystem
of some kind ? (Without having to have all of /tmp as a ramfs)
Regards,
Simon
---
This SF.net email is sponsored by
hit on only 23% of Spam.
I didn't check the overlap between the two. (Since it requires more than a
simple search/grep and I can't be bothered ;)
Regards,
Simon
---
This SF.net email is sponsored by: IBM Linux Tutorials.
Become an
eems to be the case with
2.60.
The main problem I notice with them is despite repeated manual and
automatic training on them, nigerian spams still frequently get either a
neutral bayes score (giving 0 points) and quite often a very hammy bayes
score giving them enough negative points to offset the po
messages
come in...(any server can be brought to its knees with enough messages
arriving simultaneously if you don't have sensible limits on SA
concurancy)
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Program.
Does
ately deleted. Without
the immediate delete above 15 there are just too many messages in the spam
folder to look through, with delete above 15, there are only a handfull and
I can be 99% confident nothing important scored over 15.
Regards,
Simon
--
At 14:49 24/11/2003 -0800, Ted Cabeen wrote:
Simon Byrnand <[EMAIL PROTECTED]> writes:
> I could be wrong, but I'd say that "bus error" means your system
> either ran right out of memory (including virtual memory) during the
> execution of sa-learn (and sa-learn us
A bus error can't be a SpamAssassin bug per se, or even a perl bug.
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive? Does it
help you create better
if they do, they're
more likely to scoff at the "review" and see it for what it is, rather
than be swayed by it...
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help y
At 23:01 23/11/2003 +, Alan Munday wrote:
Thanks Simon
I am using stars as I used the Advosys how-to as my starting point. I've
started playing with the scripts and wondered if SA left any variables set
e.g. hit value, that I could use rather than re-reading the file as I
thought this
: YES
is the correct way to tell if a message has exceeded the threshold...while
counting the stars is also ok for seeing if a message exceeded an arbitary
threshold. (Remember though that since 2.60 it will never add more than 50
stars regardless of the score being higher than that)
Regards,
it can't
work it out from email headers...)
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive? Does it
help you create better code? SHARE THE LOVE, and hel
ount.
In fact if you search the archives for messages from me you should find a
procmail script I posted that lets you limit the local delivery concurancy
to a manageable level. Also if you're using spamc, (and you probably should
be) look at the -m op
d or suffering from a DoS attack.
Anyone else have any ideas ?
The pyzor client itself is quite buggy too, nearly a quarter of the sample
messages I've tried manually running through it cause it to segfault, so
until they bring out a new version I think I'll probably leave it off and
stic
suspect a few others have done the same :)
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive? Does it
help you create better code? SHARE THE LOVE, and help us help
nths ago - 50% spam
1 month ago - 60% spam
last week - 63% spam
:-(
Regards,
Simon
---
This SF.Net email sponsored by: ApacheCon 2003,
16-19 November in Las Vegas. Learn firsthand the latest
developments in Apache, PHP, Perl, XML, Java, My
garding problems with
AMD's as well. (I stay clear of AMD CPU's on mission critical servers
exactly for this reason)
Good luck.
Regards,
Simon
that contain any
> of these strings:
>
> spam
> abuse
> postmaster
> .gov
> ftc
Some perhaps, but certainly not all. I regularly receive spam to abuse@
and postmaster@ which unfortunately I have to have listed in all_spam_to
in case someone
At 20:43 20/10/2003 -0500, David B Funk wrote:
On Tue, 21 Oct 2003, Simon Byrnand wrote:
> Since I use sendmail, which unfortunately doesn't have a proper way to
> limit local delivery concurancy, I'm now using a combination approach - I
> use -m 15, and return EX_TEMPFA
u can't
get hundreds of local delivery processes queued. (At most, 40)
Regards,
Simon
---
This SF.net email is sponsored by OSDN developer relations
Here's your chance to show off your extensive product knowledge
We want to know what
n.
When you say you "disabled" bayes locking, are you refering to using
'bayes_learn_to_journal 1' ?
Regards,
Simon
---
This SF.net email is sponsored by OSDN developer relations
Here's your chance to show off your exten
game over.
This general problem has been discussed *many* times on the list...
Are you using the -m option for spamd ? Are you limiting local delivery
concurancy at the MTA level ?
*ANY* server, no matter how grunty can be overwhelmed and pushed into
swapping if the incomming conne
> Simon Byrnand wrote on Thu, 16 Oct 2003 13:41:35 +1300:
>
>> 128MB is not enough in my experience. I found that with 128MB of ram
>> that I
>> had to limit concurant scanning to no more than 5 spamd processes at
>> once
>> or a burst of incomming traf
call spamc instead of
spamassassin. (It's faster to include the full path to spamc by the way)
Regards,
Simon
---
This SF.net email sponsored by: Enterprise Linux Forum Conference & Expo
The Event For Linux Datacenter Solutions & St
place to do this is in
your MTA's configuration...
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Program.
SourceForge.net hosts over 70,000 Open Source Projects.
See the people who have HELPED US provide better servi
robably need to do:
* Add more physical ram
* Limit the maximum number of spamassassin processes that can run at once
to a value that doesn't use too much memory
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Prog
ple on other lists got them, too. Mostly spamfighters, but
> not all.
Awww I didn't get one not fair ;-)
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Program.
SourceForge.net hosts over 70,000 O
At 17:55 13/10/2003 -0400, landy wrote:
On Mon, 2003-10-13 at 20:54, Simon Byrnand wrote:
> >
> > We run around 50%. And that's by count. With the MS worms flying in we
> > have noticably more spam by volume than real mail.
>
> Our current stats are 57% Spam, 43
e the world is heading when more email is junk
than legitimate :/
Regards,
Simon
---
This SF.net email is sponsored by: SF.net Giveback Program.
SourceForge.net hosts over 70,000 Open Source Projects.
See the people who have HELPED US provide b
intermediate relay server will then have to bounce
it. Usually to some innocent third-party. (Like me. I receive a few
thousand such bounces per week.)
This can be solved by running SpamAssassin on the backup MX's as well of
course(with some kind of mirr
way with various software is always read the
INSTALL file :) On a lot of software I find the INSTALL file more
informative than the README file...(which usually just tells you what the
program does, and not much about how to implement it..)
Regards,
Simon
--
0 non-token data: last expiry atime
> 0.000 0 43200 0 non-token data: last expire atime
> delta
> 0.000 0 94019 0 non-token data: last expire
> reduction count
>
> Which is saying that 281989 spam's are learn't and 9
n-token data: last expire
reduction count
Which is saying that 281989 spam's are learn't and 99372 hams..
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_
lot of open source type software, you get as much out of it as
you put into it, and one of the things that needs to be put into open
source software is the time to read all the docs thoroughly before
installing it, rather than closing your eyes, typing "make install", and
hoping for the
st ancient versions...)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk mailing list
[EMAIL PROTEC
AM** for example)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourc
do that, both razor and pyzor specifically ask that only spam
verified by a human being as spam is submitted, they specifically don't
want automated tools like SpamAssassin submitting messages without human
intervention...
Regards,
Simon
at stripping headers
than SpamAssassin's markup removing mode. (C vs Perl)
That only leaves the header change to figure out(perhaps sed, or you
could use procmails $MATCH builtin to extract the subject minus the
spamassassin change, and use the same invocation of formmail thats
stripping o
scores based on a threshold of 5.
Having said that *our* sitewide default is 7 because even with 2.55 I
notice too many false positives with 5. (Not on the kind of mail that *I*
receive, but many of our customers....)
Regards,
Simon
---
This
e new default of 0.1
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
y that teaching the message to BAYES
using sa-learn in 2.60 should be very effective, as BAYES_99 in 2.60 has a
high enough score to tag as spam without any other tests...
Regards,
Simon
---
This sf.net email
> On Thu, Sep 11, 2003 at 04:32:47PM +1200, Simon Byrnand wrote:
>> Anybody running spamc from procmail with sendmail is vulnerable to
>> having
>> their server overloaded because of uncontrolled concurancy of local
>> deliveries - despite the options mentioned above, se
> Just for a data point, rc4 did not overwrite local.cf on my linux
> machine. "perl Makefile.PL; make; make install"
Nor mine...
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http
imum. YMMV.
* Using this in combination with the spamd -m option may improve throughput
slightly by doing a limited amount of stacking of waiting procmail
processes. For example setting spamd's -m option to 10 and setting SPAMCMAX
to 15. (Untested)
* Use at your own r
e to
figure out how to fool a simple regex in a publically available program ?
Besides, whos to say no-one ever spams through Microsoft Exchange ;-)
Regards,
Simon
---
This sf.net ema
without making a donation.
Spamcop can be turned on just by setting the score to something that's not 0.
Which brings up a good question that I've been meaning to ask. If spamcop
is not free to use, why does 2.60-rc3 enable it by default ?
score RCVD_IN_BL_SPAMCOP_NET 0 2.
Essentially this would be off-by-default extra information/debugging
information.
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
unctioning properly,
and that it could be an integration problem...
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk ma
is reduce the initial communication setup
latency between spamc and spamd, and also allow large messages to be
transfered through to spamd and back again faster, but the actual scanning
of the message would be the same as before. As this is likely to be the
bulk of the scanning time I
> Simon Byrnand wrote:
>
>> At 20:43 4/09/2003 +0200, Ralf G. R. Bergs wrote:
>>
>>> Jim Porter wrote:
>>>
>>>> score RCVD_IN_OSIRUSOFT_COM 0
>>>> score X_OSIRU_DUL 0
>>>> score X_OSIRU_DUL_FH0
the score of an RBL check to zero
*does* disable the test itself, and has be confirmed by the developers on
previous occasions
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven
es not to be learnt.
There are quite a number of reason for autolearning not to learn something,
its not just a matter of if its greater than 15 learn it
Regards,
Simon
---
This sf.net email is sponsored
At 14:20 2/09/2003 +0200, Carlo Wood wrote:
On Tue, Sep 02, 2003 at 01:02:47PM +1200, Simon Byrnand wrote:
> Well don't forget that the auto_learn_spam threshold is 15 in 2.55 and 12
> in 2.60, and its very rare that a spam pasted into the *body* of a message
> will be autolearnt,
ay to
find this out from spamd directly rather than using ps ?) and if its over a
set limit, exit with EX_TEMPFAIL. A workaround like that could be used by
anyone calling spamc from procmail...
I might have a tutu around later and see if I can work out a procmail
recipie to d
a limit on concurant
scanning, however it will also cause connections to be rejected when this
limit is reached. (As well as that, 24 simultaneous spamassassin processes
is enough to bring most machines to their knees)
Regards,
Simon
---
This sf.ne
What you say is true of course...(as well as funny)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk mailing l
te normal for example for spamd to use close to 20MB
of ram.
Also, how are you calling spamassassin. (spamc, spamassassin, or a direct
filter plugin etc)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven
At 06:02 2/09/2003 +0100, Justin Mason wrote:
On Tue, Sep 02, 2003 at 01:02:47PM +1200, Simon Byrnand wrote:
> At 02:24 2/09/2003 +0200, Carlo Wood wrote:
> >On Tue, Sep 02, 2003 at 09:56:54AM +1200, Simon Byrnand wrote:
> I don't whitelist this mailing list and I know at
At 02:24 2/09/2003 +0200, Carlo Wood wrote:
On Tue, Sep 02, 2003 at 09:56:54AM +1200, Simon Byrnand wrote:
> On the other hand, there is nothing to stop the message being
autolearnt if
> its score before the whitelisting value is added, so for example if a spam
> would normally score
At 17:50 1/09/2003 +0200, Carlo Wood wrote:
On Mon, Sep 01, 2003 at 04:31:34PM +0200, Kai Schaetzl wrote:
> > whitelisted mails should not be auto learned.
>
> That is just what Simon said and SA does.
Oh... sorry. He said
S> Please check the docs, Bayes auto_learn does not t
s with tflags set to 'userconf' (user
white/black-listing rules, etc)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamass
la ticket
in about it just in case it gets forgotten :)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk mailing li
otherwise it won't learn ham.
>
> --Luke
Hi,
Did you actually check what the default is for 2.60 ? In rc2 and rc3 the
default is 0.1 for ham and 12 for spam... (we should give the developers
some credit ;-)
Regards,
Simon
>
>
> As
>"Matt Kettler" <[EMAIL
> On Friday 29 August 2003 04:34 CET Theo Van Dinter wrote:
>> On Fri, Aug 29, 2003 at 01:57:27PM +1200, Simon Byrnand wrote:
>> > Fixed now, but the second part of Theo's fix (assuming he did it :)
>> > doesn't seem to be in there - using any \n's to add
> Installed fine, upgraded fine from cvs of three weeks ago or so. Can't
> comment on effectiveness yet. One thing I note is that it again needs more
> RAM, it's now at almost 25 MB.
Not for me it size, Size of 19996, and RSS of 17M after running for a
while... the same as 2.
sin or /usr/local/etc/mail/spamassassin), and use that
> one. Or, ask, like it does for Razor2.
>
> Simply switching between versions however is a bad idea.
I agree wholeheartedly... There are enough surprises for the unwary
between 2.55 and 2.60 alr
l/etc to
> /etc).
> See also bug 2374 [1].
Argh, more traps...
So what is the behaviour when running perl Makefile.PL with no command
line arguments ? Will it be looking for /etc/mail/spamassassin/local.cf ??
Regards,
Simon
---
This sf.net
> Simon Byrnand <[EMAIL PROTECTED]> writes:
>
>> So umm, how does the rbl_timeout setting work in 2.60 then ? I didn't
>> quite
>> follow the logic of what you said :) I would have previously assumed
>> that
>> it was just a cutoff where if an ind
At 22:38 28/08/2003 -0700, Daniel Quinlan wrote:
Simon Byrnand <[EMAIL PROTECTED]> writes:
> Although I don't have the answer to your question, I suggest you look at
> using the following options to reduce the various timeouts to minimize the
> chance of a "train wreck&qu
ables automatic header folding causing long lines...
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk mailing list
[
iling when they shouldn't...
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https
> On Thu, Aug 28, 2003 at 10:31:50PM +1200, Simon Byrnand wrote:
>> RBL checks are on by default, but you have to have the Net::DNS perl
>> module installed for them to work.
>
> Woah!
>
> I DON'T have that installed! And I never got a warning anywhere?!
>
is through
CPAN.
perl -MCPAN -e shell
install Net::DNS
quit
After that you want to try running a message through SpamAssassin manually
from the command prompt in debugging mode and look at the debugging output
for mention of DNS and RBL che
rusoft tests for 2.60-rc3, so by the time 2.60 comes out the tests will
be gone.
It's also on the news page of the website how to disable the tests for
2.55, and its been discussed at least 4 or 5 times already in the mailing
list, so theres not much more that can be done...
Regards,
Simon
ll be a bit of short term pain (for some people) for longterm
gain :)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Sp
At 18:41 27/08/2003 -0700, Justin Mason wrote:
Simon Byrnand writes:
>Just got a sales pitch today by phone followed up by email from these guys:
>
>http://www.death2spam.net.nz/
>
>Anybody else heard of them ? Their system claims to be based on Bayesian
>filtering and claim
tm) adds a whole host
of new RBL checks. Unless you have a specific RBL you're interested in that
isn't currently supported by SpamAssassin, then it is *already* using a
selection of RBL lists...
Regards,
Simon
Selon Bob Apthorpe <[EMAIL PROTECTED]>:
> On Wed, 27 Aug 200
At 18:41 27/08/2003 -0700, Justin Mason wrote:
Simon Byrnand writes:
>Just got a sales pitch today by phone followed up by email from these guys:
>
>http://www.death2spam.net.nz/
>
>Anybody else heard of them ? Their system claims to be based on Bayesian
>filtering and claim
At 20:51 27/08/2003 -0400, Theo Van Dinter wrote:
On Thu, Aug 28, 2003 at 12:40:12PM +1200, Simon Byrnand wrote:
> I don't think I'll be switching away from SpamAssassin any time soon...
>
> Comments anyone ?
98% of all statistics are lies. :)
Heheheh... I don't think I
should ignore it. Well
duh! Spam anyone ? :) (And yes, their bogus virus reports are detected by
both DCC and Razor now)
Regards
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
s...
Oh yeah, it claims to be written in Java too.. :)
I don't think I'll be switching away from SpamAssassin any time soon...
Comments anyone ?
Regards,
Simon
---
This sf.net email is sponsored
razor and DCC :)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourc
d showed
> no SA checks. I too reverted back to 2.55.
Have you tried rc2 ? I read that it was a bug in the DNS lookups that
caused the message to fail to be scanned if one of the DNS requests timed
out... Aparently fixed in rc2. (I havn't seen that problem o
> By all means do it! :) ... but it won't help that there are still
> servers
> that use the rbl directly from sendmail or postfix and the like.. :(
And they will soon discover that they've been rejecting nearlly all mail
because of it, and remove relays.osirusoft.com.
>
> Simon Byrnand writes:
>> I was just thinking about the GA process and although I havn't looked at
>> it to see exactly how it works, I was wondering the following
>>
>> Presumably it starts with a certain scoreset, runs the spam through,
>> sees
&g
Good reason to get 2.60rc3 out ASAP ;)
So are the scores just zero'ed out for rc3 then ? I suppose its not worth
the effort and time of another GA run...:)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkGeek
Welcome
ore X_OSIRU_DUL 0
score X_OSIRU_DUL_FH 0
score X_OSIRU_OPEN_RELAY 0
score X_OSIRU_SPAMWARE_SITE 0
score X_OSIRU_SPAM_SRC 0
(Hope I didn't miss any)
Regards,
Simon
---
This sf.net email is sponsored by:ThinkG
s for, I think there would be a
defintate improvement in FN/FP rate of an independant corpus using the 6/4
threshold instead of 5/5.
Anybody able to blow holes in my theory or suggest a way of proving it ?
Regards,
Simon
---
This sf.net email is spon
1 - 100 of 356 matches
Mail list logo
