[My e-mail address in gcc-bugs mailing list archive]

[My e-mail address in gcc-bugs mailing list archive]

[Chris Faylor]

Maybe we need some kind of general admonition against this?
It really isn't nice to include someone else's raw email addresses
in the text of a message.

----- Forwarded message from Thomas Henlich <henlich<*>mmers1<^>mw<^>tu-dresden<^>de> -----

Dear maintainer of the gcc-bugs mailing list archive,

this message http://gcc.gnu.org/ml/gcc-bugs/2001-03/msg00965.html
contains my full e-mail address WITHOUT your usual transformations.

To fight spam and protect my privacy, I am asking you:

Please could you make the same transformations that are usually done for
the e-mail headers (e. g. replacing it with henlich at mmers1 dot etc).

Thank you in advance,
 Thomas.

PS: Keep up the good work!

----- End forwarded message -----

Re: [My e-mail address in gcc-bugs mailing list archive]

[Ian Lance Taylor]

Chris Faylor <me@cgf.cx> writes:

> Maybe we need some kind of general admonition against this?
> It really isn't nice to include someone else's raw email addresses
> in the text of a message.

No, but in our context it's nearly impossible to avoid, when people
start quoting ChangeLog entries.

Ian

Re: [My e-mail address in gcc-bugs mailing list archive]

[Christopher Faylor]

On Mon, Feb 28, 2005 at 12:27:59PM -0500, Ian Lance Taylor wrote:
>Chris Faylor <me-SMACK-cgf-PERIOD-cx> writes:
>
>> Maybe we need some kind of general admonition against this?
>> It really isn't nice to include someone else's raw email addresses
>> in the text of a message.
>
>No, but in our context it's nearly impossible to avoid, when people
>start quoting ChangeLog entries.

Right, but there are certain conventions which we could avoid, like,
for instance, the above, where you quoted my raw email address in
the "... writes:"

I know that this is basically a losing battle but I don't see any reason
to make things even slightly easier for spammers.

I once tried to come up with a general filter which could be applied to
email messages so that ChangeLogs were left alone but anything after a >
or a "From: " was munged.  Maybe I should resurrect that.

cgf

Re: [My e-mail address in gcc-bugs mailing list archive]

[Ian Lance Taylor]

Christopher Faylor <who-has-no-email-address-especially-not@cgf.cx> writes:

> On Mon, Feb 28, 2005 at 12:27:59PM -0500, Ian Lance Taylor wrote:
> >Chris Faylor writes:
> >
> >> Maybe we need some kind of general admonition against this?
> >> It really isn't nice to include someone else's raw email addresses
> >> in the text of a message.
> >
> >No, but in our context it's nearly impossible to avoid, when people
> >start quoting ChangeLog entries.
> 
> Right, but there are certain conventions which we could avoid, like,
> for instance, the above, where you quoted my raw email address in
> the "... writes:"
> 
> I know that this is basically a losing battle but I don't see any reason
> to make things even slightly easier for spammers.

I would have to agree with the "losing battle" part.

> I once tried to come up with a general filter which could be applied to
> email messages so that ChangeLogs were left alone but anything after a >
> or a "From: " was munged.  Maybe I should resurrect that.

Sounds like a good idea.  In fact, I think it would be fine to munge
ChangeLog entries too.

Ian

Re: [My e-mail address in gcc-bugs mailing list archive]

[Christopher Faylor]

On Mon, Feb 28, 2005 at 12:37:09PM -0500, Ian Lance Taylor wrote:
>Christopher Faylor <who-has-no-email-address-especially-not@cgf.cx> writes:

LOL!

>> On Mon, Feb 28, 2005 at 12:27:59PM -0500, Ian Lance Taylor wrote:
>> >Chris Faylor writes:
>> >
>> >> Maybe we need some kind of general admonition against this?
>> >> It really isn't nice to include someone else's raw email addresses
>> >> in the text of a message.
>> >
>> >No, but in our context it's nearly impossible to avoid, when people
>> >start quoting ChangeLog entries.
>> 
>> Right, but there are certain conventions which we could avoid, like,
>> for instance, the above, where you quoted my raw email address in
>> the "... writes:"
>> 
>> I know that this is basically a losing battle but I don't see any reason
>> to make things even slightly easier for spammers.
>
>I would have to agree with the "losing battle" part.
>
>> I once tried to come up with a general filter which could be applied to
>> email messages so that ChangeLogs were left alone but anything after a >
>> or a "From: " was munged.  Maybe I should resurrect that.
>
>Sounds like a good idea.  In fact, I think it would be fine to munge
>ChangeLog entries too.

Well, if it's ok to munge ChangeLog entries, then we could munge all email
addresses, right?

The only reason I haven't pursued this is because I hate to sacrifice
convenience for the sake of avoiding evil.

Actually, if we munged everything consistently, we could provide an interface
which gave you raw email addresses again, if you knew the secret handshake.

Perhaps Jason Molenda would like to comment on this interesting new idea that
I have just invented now, off the top of my head, without any prior knowledge
of anything which could potentially have been done before...

cgf

Re: [My e-mail address in gcc-bugs mailing list archive]

[Jason Molenda]

Hi all, sorry for not following the discussion too closely...

Yeah, I know mhonarc can be configured to do all sorts of munging
on the contents of messages.  I haven't looked at it recently (I've
been meaning to update the version on sourceware for a year or two
now... sigh), but we're not treading new ground here.


On Mon, Feb 28, 2005 at 12:44:06PM -0500, Christopher Faylor wrote:

> Actually, if we munged everything consistently, we could provide an interface
> which gave you raw email addresses again, if you knew the secret handshake.
> 
> Perhaps Jason Molenda would like to comment on this interesting new idea that
> I have just invented now, off the top of my head, without any prior knowledge
> of anything which could potentially have been done before...


Hehe, yeah, as Chris implies, the "get raw text" cgi-mechanism that
all the mailing list archives use does its munging on the fly; the
files on disk are stored unmunged.  Right now it only munges addresses
in headers; it would have to be modified by hand to munge addresses
in the body of messages.

But Chris, I'm not sure what you're impling here?  An option to the cgi
script that would NOT munge the headers?  Surely that would be exploitable
by spammers, wouldn't it?  Sounds like security through obscurity to me.
Not a good plan, that.  nomunge=1.

J

Re: [My e-mail address in gcc-bugs mailing list archive]

[Christopher Faylor]

On Mon, Feb 28, 2005 at 09:57:59AM -0800, Jason Molenda wrote:
>Hi all, sorry for not following the discussion too closely...
>
>Yeah, I know mhonarc can be configured to do all sorts of munging on
>the contents of messages.  I haven't looked at it recently (I've been
>meaning to update the version on sourceware for a year or two now...
>sigh), but we're not treading new ground here.
>
>On Mon, Feb 28, 2005 at 12:44:06PM -0500, Christopher Faylor wrote:
>
>>Actually, if we munged everything consistently, we could provide an
>>interface which gave you raw email addresses again, if you knew the
>>secret handshake.
>>
>>Perhaps Jason Molenda would like to comment on this interesting new
>>idea that I have just invented now, off the top of my head, without any
>>prior knowledge of anything which could potentially have been done
>>before...
>
>Hehe, yeah, as Chris implies, the "get raw text" cgi-mechanism that all
>the mailing list archives use does its munging on the fly; the files on
>disk are stored unmunged.  Right now it only munges addresses in
>headers; it would have to be modified by hand to munge addresses in the
>body of messages.
>
>But Chris, I'm not sure what you're impling here?  An option to the cgi
>script that would NOT munge the headers?  Surely that would be
>exploitable by spammers, wouldn't it?  Sounds like security through
>obscurity to me.  Not a good plan, that.  nomunge=1.

I wasn't exactly sure what I was implying either.  I guess it was
something like a "nomunge=1" option, the theory being that spammers
wouldn't be following the gcc mailing list where this setting could be
announced.

It is security-through-obscurity but it's similar to the current spam
blocking system.  A spammer just has to subscribe themselves to the
global-allow list if they want to spam mailing lists but I haven't seen
any clear indication of anyone doing that yet.  Having to special case
something like this seems to be contrary "send millions of emails to get
one response" economy that spammers use.  If they have to do research
to figure out how to spam or scrape this one mailing list, it doesn't
seem like that will be a very common thing.

Even if it was common, it won't be any worse than what we have now and
it should be slightly better.

(of course, I'm sure that all of this has been rabidly discussed on some
spam-related mailing list or newsgroup somewhere)

cgf

Re: [My e-mail address in gcc-bugs mailing list archive]

[Jason Molenda]

On Mon, Feb 28, 2005 at 01:05:04PM -0500, Christopher Faylor wrote:

> I wasn't exactly sure what I was implying either.  I guess it was
> something like a "nomunge=1" option, the theory being that spammers
> wouldn't be following the gcc mailing list where this setting could be
> announced.


That makes good sense.  Not only would they have to read the list,
but they'd have to modify their crawler-bot to add nomunge=1 followed
by an ampersand option-delimiter in the URL of the get raw message.
I don't think we're a juicy enough target for the spambot-authors to
be monitoring how sourceware is set up and modify their software
specifically for us.

FWIW, all joking aside, my main concern with munging the bodies of
messages was that patches could fail to apply if there were e-mail
addresses in comments.  If we munge e-mail addresses in msg bodies
of both the mhonarc and get-raw-text (without an option like "nomunge=1")
accessor methods, people will have to modify those patches by hand.
Incidentally, I don't know if mhonarc allows for munging in
attachments, e.g. patches attached to an e-mail message.

J