This document is available online at http://www.c2.net/~raph/premail/.

Introduction to premail

This is the documentation for version 0.45 of premail, an e-mail privacy package by Raph Levien. It is organized as a single, large document so as to be easily readable when printed. You can, however, jump directly to one of these topics: installation, secrets, preferences, Netscape, Pine, other mailers, command line, encryption, decoding, anonymity, nyms, usenet, address book, smime, debugging, technical notes, related documents, (end of list).

The main function of premail is adding support for encrypted e-mail to your mailer, using plain PGP, PGP/MIME, MOSS, or S/MIME.

In addition, premail provides a seamless, transparent interface to the anonymous remailers, including full support for Mixmaster remailers and the nymservers. Nymservers provide cryptographically protected, fully anonymous accounts for both sending and receiving e-mail.

While premail can be used as a stand-alone application, it works best when integrated with your mailer. Currently, premail is integrated completely seamlessly and transparently only with Netscape 3.0's built-in mailer. It works fairly well with Pine 3.94 or later, as well (plain PGP is supported, but decryption of MIME-based e-mail encryption protocols is still missing). Transparent integration of outgoing mail only is supported for any mailer in which the mail sending program can be configured, including Berkeley mail, most emacs mailers, and MH. For these mailers, you can decode messages with a single command.

To integrate with your mailer, premail places itself between the mailer and the actual mail transport. For outgoing mail, premail masquerades as sendmail. You configure your mailer to call premail instead of sendmail. Then, premail performs the encryption or signing, and invokes sendmail to actually send the message.

For mailers that call a command to receive incoming mail (including Netscape 3.0), the situation is similar. Netscape, for example, can be configured to call movemail to get incoming mail. To integrate premail, you'd configure Netscape to call premail instead, which would in turn call movemail to actually get the mail, then would decode it.

You need the following software in order to effectively use premail:

• Unix. Unfortunarely, premail does not work on Mac or Windows.
• Perl 5.000 or later.
• PGP (version 2.6.2 recommended).
• RIPEM 3.0b2 or later (optional, for S/MIME support)
• TIS/MOSS 7.1 (optional, for MOSS support)
• Mixmaster (optional, for higher security anonymous mail)
• Lynx (only if you're behind a firewall)

First, you need to get premail. The source code is available from an export-control Web server. You may also be able to find a copy on the Hacktic FTP site in the Netherlands. In either case, you want to get the file premail-0.45.tar.gz.

After you've gotten the file, unpack it. This command should do it:

   gzip -dc premail-0.45.tar.gz | tar xvf -

The unpacking process will create a subdirectory called premail-0.45, containing the following files:

README	A short description of the contents
premail	The premail program itself
preferences	A skeletal preferences file

Test to see if you can run premail. These commands should print a usage summary:

   cd premail-0.45
   ./premail

If you get an error message reading "command not found," then you will have to edit the first line of premail to refer to the actual pathname of the perl5 interpreter. One good way to find out the pathname is to do "which perl5" or "which perl".

On the other hand, if you get a string of syntax errors, then the problem is that you are running perl4, while premail needs perl5. Try to see if you can find perl5 on your machine. Otherwise, you may need to install perl5 yourself.

If you will be using premail from the command line frequently, then you may want to copy (or symlink) the premail program into a location in your $PATH. For example, if you have permission to add files into /usr/local/bin, then you may consider running this command:

   cp -p premail /usr/local/bin

At this point, you are ready to test whether premail actually works. We are assuming that you already have PGP installed and have generated your own public key. Type this command, substituting in your own e-mail address:

   ./premail -t
   To: your@own.email.addr ((encrypt-pgp))
   Subject: Test

   Does this really work?
   .

If all goes well, you should be back at the command line within a couple of seconds. If it seems to hang without any disk or net activity, try typing randomly for a minute, under the assumption that PGP needs random keystrokes. This shouldn't happen if PGP is already set up correctly (including having generated your own public key), but on the chance that it isn't, hanging while waiting for random keystrokes is one of the more common failure modes.

This is also the point at which you may get a PGP error. Two common problems are that premail can't find the PGP program, in which case you will want to add a line to your preferences file (see below), or that it can't find the public key corresponding to your e-mail address.

If the test was successful you now have a PGP-encrypted message in your mailbox, then you should now have a PGP-encrypted message in your mailbox.

While premail's default configuration is designed to be sufficient for the the most common cases, you may want to change some of the configuration options. This is done by adding lines to the preferences file.

The default location for the preferences file is ~/.premail/preferences, where ~ represents your home directory. The premail distribution comes with a skeleton preferences file, but it does not automatically copy it into the ~/.premail directory. You might choose to do that yourself, or you might create one from scratch.

The format of the preferences file is a sequence of lines such as the following:

   $config{'option'} = 'value';

All other lines (including those beginning with #) are considered to be comments and are ignored. Here's a typical preferences file (actually, the one on my home machine):

   $config{'logfile'} = '/home/raph/premail/log';
   $config{'debug'} = 'chvl';
   $config{'movemail'} = '/home/raph/bin/movehome';
   $config{'ripem'} = '/home/raph/install/ripem/main/ripem';
   $config{'pgp'} = '/usr/local/bin/pgp';

As you can see, a major use for the preferences file is to specify full pathnames for the helper programs. In addition, I've set it up to produce a full log, which I find useful, because I'm constantly tracking down bugs :-)

Here's a table of all the configuration options, their defaults, and a very brief description. More complete descriptions are found in the preferences file included in the premail distribution.

To create signatures, decrypt messages, or use nyms, you need to set up a "premail secrets" file. If you will only be using premail to encrypt outgoing mail, you can skip this section.

The default filename is /tmp/.premail-secrets.$< , where $< is equal to your numeric user id. To change the filename, use a preferences line such as this one:

   $config{'premail-secrets'} = '/mnt/cryptdisk/premail-secrets';

If you don't know your numeric user id, you can find it by running "echo $uid" (from csh or tcsh), "echo $UID" (from sh or bash), or:

   perl -e 'print "$<\n"'

The premail secrets file has this format:

   $pgppass{'user'} = 'PGP passphrase for user';
   $pgppass{'alternate'} = 'PGP passphrase for alternate';
   $penetpass = 'Passphrase for anon.penet.fi';

However, make sure your premail secrets file has restrictive permissions, so other people on your system can't read your passphrases! This command is well recommended (substituting your actual user id, of course):

   chmod 600 /tmp/.premail-secrets.7437

Logging in and logging out

Generally, premail stores its secrets file in the /tmp directory. In some cases, this is good enough security. In other cases, it might be better to store the file encrypted most of the time, and only decrypt it when necessary. To use this capability of premail, first set a passphrase with:

   premail -setpass

You will be prompted for a passphrase. You can use the same passphrase as for your PGP key, or a different one, depending on how many passphrases you want to remember. This command leaves you logged in with the new passphrase set.

To log out:

   premail -logout

You might consider adding this command to your .logout file, so that it occurs automatically every time you log out of your account.

To log in again:

   premail -login

If you are running on a system with X, then premail will automatically pop up a window to log in whenever the secrets are needed. If you are not running X, and the secrets are needed, you will get an error. In this case, you can log in manually and try the command again.

This section describes how to integrate premail into Netscape 3.0's built-in mailer. Skip this section if you won't be using Netscape mail.

1. Create symbolic links to premail called "prezilla" and "premailmove". To do this, make sure you are in the same directory as premail itself, and type:

   ln -s premail prezilla
   ln -s premail premailmove

2. Find a working movemail. If you have emacs installed, then you most likely have one in /usr/lib/emacs/etc/movemail or a similar location. If you don't already have one, then the source (or possibly binary) for one is included in the Netscape Navigator distribution and you can build it (no need if a binary is included). Then, make sure premail can find it by adding a line such as this one to your preferences file:

   $config{'movemail'} = '/usr/lib/emacs/etc/movemail';

This usage assumes that you get your mail from a mail spool, as opposed to POP or some such. You may be able to get it to work for POP as well, but you need to figure out how to invoke movemail to move the mail from your mailbox to a file (specified as the second argument to the movemail script).

3. Add this line to your .cshrc, assuming your shell is csh or tcsh:

   setenv NS_MSG_DELIVERY_HOOK /your/path/to/prezilla

Also run this command from the shell so it takes effect immediately. The syntax is slightly different if your shell is sh or bash (note: is this right?):

   NS_MSG_DELIVERY_HOOK=/your/path/to/prezilla
   export NS_MSG_DELIVERY_HOOK

4. Start Netscape (exit first if it's already running). Go to the Options|Mail and News Preferences dialog, select the Servers tab. Click on "External Movemail" and set the value to /your/path/to/premailmove.

Try sending yourself mail, and clicking on "Get Mail" from the Netscape Mail window. The mail should show up in the Inbox, correctly decoded.

To view the X-Premail-Auth: header field to see the result of signature checking, select Options|Show All Headers from the Netscape Mail window.

Note: as of Netscape v3.0, there is still a bug in the handling of the Bcc: header field, which causes it to be ignored. Do not use this field. Hopefully, this will be fixed in a future version of Netscape.

Note: some 3.0 beta versions modify the PATH environment variable. If premail seems to work correctly from the command line, but not from Netscape, try setting absolute pathnames for the programs used by premail.

As of Pine 3.94, premail integrates both outgoing mail and the decryption of plain PGP incoming mail. Unfortunately, decryption of MIME-based mail is not yet supported.

Two Pine configuration options need to be set to integrate premail (i.e. from the main Pine screen, S for setup, then C for configure). First, sendmail-path should be set to a value similar to this (substituting the actual path to premail):

   /your/path/to/premail -oem -t -oi

Second, display_filters should be set to a value similar to this:

   _BEGINNING("-----BEGIN PGP")_ /your/path/to/premail -decode -body

If you have trouble finding these options in the setup screen, then you can edit the .pinerc file directly.

One caveat when using Pine: it usually tries to be "smart" and remove comments from e-mail addresses, which includes the double-paren commands such as ((encrypt-pgp)). There are a few ways to deal with this problem:

• Use "( )" instead of (( )). Note: I think this works, but I haven't tested it.
• Use the alternative caret syntax. These two lines mean the same thing:

     To: raph@cs.berkeley.edu ((encrypt-key, sign))
     To: raph@cs.berkeley.edu^encrypt-key^sign
  

• Avoid setting the encryption options on the command line altogether, and set them in the addresses file instead (see below).

This section describes how to integrate premail with MH, emacs, and UCBMail. With these mailers, premail will only handle outgoing mail automatically. To decode incoming mail, you still need to invoke premail -decode by hand.

Integrating premail with Emacs

To add premail support to emacs, just add this line to your .emacs file:

   (setq sendmail-program "/your/path/to/premail")

Integrating premail with MH

In whatever directory you keep the premail executable, create a symbolic link as follows:

   ln -s premail prepost

Under the name "prepost", premail will masquerade as MH's post program rather than sendmail. You can get MH to call premail instead of post by adding this line to your .mh_profile:

   postproc: /your/path/to/prepost

One thing to keep in mind is that premail's processing is done before that of post. Thus, if you have MH aliases, they will get expanded after the call to premail. If you use only premail aliases, only MH aliases, or neither, this won't be a problem.

Alternatively, if you have appropriate privileges, you can add this line to /usr/lib/mh/mtstailor:

   sendmail: /your/path/to/premail

You may also have to configure MH to call sendmail locally rather than connecting to an SMTP server. Don't do both the mtstailor and mh_profile methods -- that would run premail twice.

Installing premail with UCBmail

UCBmail is a simple mailer front-end (also known as Mail and mailx). If, when you type "mail user@site.dom", the mailer asks you for a "Subject: " line, you are undoubtedly using UCBmail. If so, you are in luck - it integrates very easily with premail. Just add this line to your ~/.mailrc file:

   set sendmail=/your/path/to/premail

Using premail with UCBmail is not very different from using premail by itself, but you do get some handy features, such as including files and using an editor on the mail.

Hopefully, you have integrated premail into your mail client, and you won't have to invoke it from the command line. However, there may still be times when it is convenient to use premail from the command line.

The most basic use of premail is as a replacement for sendmail. For example, you can send mail directly from the command line, as follows (here, the > represents the Unix prompt):

   > premail -t
   To: raph@cs.berkeley.edu ((sign))
   Subject: premail bug report

   Here's a bug in premail: ...
   .
   >

The -t option specifies that the recipients are extracted from the header fields (To:, Cc:, Bcc:, and the Resent- variants of each). As in sendmail, you can specify the recipients on the command line instead of using the -t option.

In addition, you can set configuration options from the command line, using the +option=value syntax. This is especially useful with the debug option. For example, to show you what happens when formatting mail for remailers, but not actually send the message:

   > premail +debug=ry -t
   To: raph@cs.berkeley.edu ((chain=1))
   Subject: test of remailer

   test
   .
   Chose chain exon
   /usr/lib/sendmail -oi remailer\@remailer\.nl\.com << -eof-
   To: remailer@remailer.nl.com

   ::
   Encrypted: PGP

   -----BEGIN PGP MESSAGE----- remailer@remailer.nl.com
   ::
   Request-Remailing-To: raph@cs.berkeley.edu

   ##
   Subject: test of remailer

   test
   -----END PGP MESSAGE-----
   -eof-

There is one configuration option that can only be set from the command line in this fashion, which is the location of the preferences file itself. The configuration option is preferences, and the default value is ~/.premail/preferences.

Once you've got premail set up, actually using encryption is easy. You simply add commands in double parentheses to the e-mail addresses. The encrypt-pgp command (which can be abbreviated to key) adds encryption to the outgoing mail, and the sign command signs it.

For example, to send me encrypted mail, you'd send it to raph@cs.berkeley.edu ((encrypt-pgp)). You need to have a key with this user id on your PGP public keyring, otherwise you'll get an error message. If the user id on the key doesn't match the e-mail address, you can specify it directly. For example, to send mail directly to my workstation, but using the same public key as above, use raph@kiwi.cs.berkeley.edu ((key=raph@cs.berkeley.edu)).

Signing works much the same way. I can sign mail by adding ((sign=raph@cs.berkeley.edu)) to the outgoing address. Actually, because I set the signuser configuration option in my preferences file, all I have to add is ((sign)).

Doing both encryption and signing is just as easy. For example, to send me signed, encrypted mail, use this line:

   To: raph@cs.berkeley.edu ((encrypt-pgp, sign))

Each recipient is treated separately - the double-paren commands after an e-mail address apply to that recipient only. However, you can add a Sign: header field to indicate that your message is signed for all recipients. Example:

   To: vp@company, secretary@company, employees@company,
       friend@outside ((encrypt-pgp))
   Subject: Important announcement
   Sign:

   ...

In this example, all recipients will get a signed message, and the message to friend@outside will be encrypted as well.

The basic way to decode encrypted messages is to use premail -decode as a command line. You can either give a filename as an argument, or premail will accept the encrypted message on its standard input. In either case, the decoded message will be printed on the standard output.

The message can be a standard e-mail message (RFC 822 format), or it can be an entire mailbox. In the latter case, premail will decode each of the messages individually. If you don't have premail directly integrated into your mailer, then here's a handy way to view your mail:

   premail -decode $MAIL | more

If the message is actually encrypted, then premail will need to access the secrets file. If you are logged out of premail, then premail will try to open an xterm window for you to type the passphrase for the secrets file. If that doesn't succeed, premail will print an error message. At that point, you might choose to log in (i.e. premail -login) and then try the decoding again.

If, as in many mailers, you have easy access to the body of the message but not the header, then you can use premail -decode -body on the body. This works well for plain PGP encrypted messages, but unfortunately does not work for MIME-based message formats, because important information is contained in the header.

The results of the decoding (including signature verification) are given in an X-Premail-Auth: header field. This header field is protected against forgery; if the original message contains it, it is changed to X-Attempted-Auth-Forgery.

The original reason for writing premail was to provide good support for anonymous remailers. If you're not interested in sending anonymous mail, you can skip this section.

Sending anonymous mail is very similar to sending encrypted mail. Simply add the ((chain)) command to the recipient's e-mail address. Alternatively, you can add a Chain: header field, and the mail will be send anonymously to all recipients.

Even though the chain command is simple, a lot is going on under the surface. The default chain is 3, which asks that three "good" remailers be chosen randomly. To make sure that it makes its choice based on fresh, up-to-date information, premail downloads the remailer list and a set of PGP public keys for the remailers from the Web (the actual URLs are configuration options). After choosing the remailers, the message is multiply encrypted with the PGP public keys, and finally sent to the first remailer in the chain.

The automatic chain selection process is very good. My tests indicate that reliability is consistently above 99%. Further, the chain selection process avoids some potential problems. For example, some remailers are known not to work well in chains, probably because of incorrectly configured "block lists." Also, some remailers are "linked," in the sense of being hosted on the same machine, or being administered by the same person. Choosing a sequence of linked remailers wouldn't offer much security, so premail doesn't.

You can also choose the chain length. A shorter chain will be faster and more reliable, but less secure, and conversely for longer chains. For example, ((chain=5)) selects a chain of five remailers.

If this isn't enough control, you can specify the exact chain of remailers by hand. For example, ((chain=replay;jam;exon)) bounces the message around a few times outside the US.

Mixmaster chains are specified inside an additional set of parentheses. At the moment, there is no way to automatically select a chain of Mixmaster remailers, so you have to do it by hand. For example: ((chain=(replay;ecafe-mix;lcs))). You can even mix Mixmaster and type-1 remailers; for example, ((chain=(anon);1;(replay))) will sandwich one well-chosen remailer between the two Mixmaster remailers.

Extra header fields can be placed in the outgoing message by prefixing the header with "Anon-". A particularly common usage is an Anon-Reply-To: field, which specifies a reply-to address in the mail delivered to the recipient. The Reply-To: header field is used often enough that premail includes a default-reply-to configuration option, which automatically adds it to all anonymous messages.

The following header fields are passed through to the anonymized message, even without the Anon- prefix:

   Mime-Version:
   Content-Type:
   Content-Transfer-Encoding:
   Newsgroups:
   X-Anon-To:
   In-Reply-To:
   References:

This section describes how to create and use nyms, which are accounts for sending and receiving anonymous mail. There are two types of nymservers: alpha (named after the now defunct alpha.c2.org), and newnym. For the most part, the operation of the two is similar.

To create a new nym, type

   premail -makenym

and follow the prompts. This command is also good for updating an existing nym, which is important if one of the nym's remailers goes down.

You can also create or update a nym from the command line, as follows:

   premail -makenym you@alias.cyberpass.net your@real.email.address

When premail creates a nym, it chooses random passphrases (one for each remailer in the chain). The passphrases and other details of the nym are stored in the premail secrets file. Thus, the nym is fairly secure (much more so than, say, anon.penet.fi).

The decode mechanism handles responses to nyms, again looking up the passphrases in the premail secrets file.

You can also send mail from your nym, in one of two ways. Assume for the sake of example that your nym is you@alias.cyberpass.net. Then, you would use a chain of 2;cyber=you. Alternatively, you can use a chain of 2;cyber and include this header field:

   Anon-From: you@alias.cyberpass.net (You Know Who)

If you want the nymserver to send you a confirmation every time you send mail from your nym, add a $config{'ack'} = 'yes'; line to your preferences file.

To delete a nym:

   premail -makenym you@alias.cyberpass delete

Please delete nyms if you are not actually using them; this helps free up disk space and prevents the nymservers from being overloaded.

As of version 0.45, premail now supports the newnym type of nymserver. This nymserver is more richly featured than the alpha type. You do have to answer a few more prompts when creating nyms for the newnym type, including creating a new PGP key. It's worth it, though. The newnym servers seem to be working a lot better than the alpha ones ever did. For more information on newnym, see the nym.alias.net homepage. If you want to exchange nyms between premail and other programs (or a manual setup), then take a look at the -importnym and -exportnym commands, which are explained in the documentation for the patch that upgraded premail 0.44 to have newnym capability.

Even though some remailers can post directly to Usenet, premail does not support that. Thus, if you want to post to Usenet, you should use a mail-to-news gateway.

To find a working mail-to-news gateway, check Don Kitchen's list. There are two basic kinds: sites that scan the header fields, and sites that include the newsgroup in the address.

Using the address-parsing kind, to post to alt.anonymous, you'd just send mail to alt.anonymous@myriad.alias.net (assuming, of course, that myriad.alias.net is still functioning).

Using the header-scanning kind, send mail to mail2news@myriad.alias.net, and include this header field:

   Newsgroups: alt.anonymous

The header scanning kind has one advantage: you can cross-post to multiple newsgroups using one mail message.

One frequently asked question is: how can I follow up on a thread while posting anonymously? This is easy. Find the Message-Id: header field in the post you're responding to, and change it into a References: field in your outgoing mail.

Here's an example that ties it all together. Let's say you wanted to reply to this post:

   From: Edward Brian Kaufman <ebk8@columbia.edu>
   Newsgroups: alt.privacy.anon-server, alt.anonymous
   Subject: A few questions about anon posts
   Message-ID: <Pine.SUN.3.94L.960630113156@aloha.cc.columbia.edu>

   Hi,

   I'd like to know what the best/easiest way to do anon posts is and
   how to do them.  Thank you,

   Ed

To post the reply anonymously, send this mail:

   To: mail2news@myriad.alias.net ((chain))
   Cc: Edward Brian Kaufman <ebk8@columbia.edu> ((chain))
   Newsgroups: alt.privacy.anon-server, alt.anonymous
   Subject: Re: A few questions about anon posts
   References: <Pine.SUN.3.94L.960630113156@aloha.cc.columbia.edu>

   If you have a Unix machine, using premail is the best way. To find
   out how, read the manual.

Adding the extra encryption commands is not difficult, but it can be tedious and potentially error prone. Thus, premail provides an address book for specifying commands to be used with specific e-mail addresses.

For example, let's say that one of your correspondents tells you that she prefers mail to be PGP encrypted. Then, instead of typing ((encrypt-pgp)) every time you send her mail, you could add this line to your addresses file:

   her@email.address: ((encrypt-pgp))

The addresses file is usually at ~/.premail/addresses, but the location is a configurable option.

Another example was the hackerpunks mailing list (now defunct), in which all of the subscribers have alpha.c2.org nyms. Since haqr@alpha.c2.org had this line in his addresses file, he was able to post to the list with just "To: hpunks":

   hpunks: hackerpunks@alpha.c2.org ((chain=2;alpha=haqr))

An address book entry can also expand to a list of addresses. For example:

   alice: alice@crypto.com ((encrypt-pgp))
   bob: bwhite@got.net ((key=bobw@netcom.com))
   eric: eric@ecsl.org ((encrypt-pgp))

   friends: alice, bob, eric

Sending mail to friends would then do what you'd expect: send encrypted mail to each of alice, bob, and eric's full e-mail addresses.

Version 0.45 of premail contains limited support for S/MIME messages. Basic message formatting works, but there are problems with creating usable certificates, and there is still no support for an encryption algorithm interoperable with RC2. However, a few hearty souls may wish to experiment with the S/MIME functionality that is present. This section explains how to do it.

First, you must install RIPEM 3.0b2 (or later). This is available from the ripem export-controlled FTP site. You'll need to get an account on the server in order to download any of the export-controlled code - the GETTING_ACCESS file on the site explains how.

Once you have RIPEM installed (and the ripem configuration option pointing to the executable), create a public key with this command:

   premail -ripemkey

You will then be prompted for your e-mail address. Alternatively, you can give your e-mail address as a command line argument to premail -ripemkey.

After your key is created, you can send signed messages by adding the ((ssign)) command. If you send a signed message to another premail user, they will have your public key, and can send you mail, by using ((encrypt=your@user.id)).

The default encryption is Triple-DES. If the recipient can't handle it, then ((encrypt-des)) will fall back to plain DES, which most users will be able to decrypt - probably including "export" versions of S/MIME. Of course, the disadvantage of using plain DES is that any competent spy organization will also be able to decrypt the messages ;-).

Unfortunately, RIPEM 3.0b2 has some significant differences from other S/MIME implementations in the way it handles public key certificates. These prevent you from getting a VeriSign certificate you can use. It is, however, possible to accept VeriSign class 1 beta certificates by running the following (prompts and messages are in normal font, what you type is in boldface; you can find out the password by looking in the secrets file):

   > rcerts -u your@user.id
   Enter password to private key:
   E - Enable standard issuers...
   ...other choices...
     Enter choice:
   e
   ...V - VeriSign something or other...
   v
     Enter the number of months the certificate will be valid, or blank to cancel:
   12
     Enter choice:
   q

If you run into trouble with premail, it might be of value to turn on some of the debugging options. This can be done on the command line, or in the .premailrc file. In the former case, add a +debug=chvy argument to the command line. In the latter case, try:

   $config{'debug'} = 'chvy';

Here are the meanings of the debug options:

c: Print command line invocation.
h: Print headers of input message.
l: Debug output goes to log instead of stdout.
p: Print finished message, do PGP.
r: Print chain chosen (useful in debugging chain selection).
y: Print finished message, don't do PGP.
v: Print all kinds of verbose info.

Note that +debug=p puts the encrypted message on stdout. This may be useful for constructing reply blocks, among other things.

If there are problems with premail, then one of the best ways to track them down is through the log. Try setting the debug configuration option to chvl, setting the logfile configuration option (for example, to ~/.premail/log), and then examining the log. Also, if you're bringing bugs to my attention, it helps a lot if you can send me relevant excerpts from the log.

This section covers a number of techincal notes related to the operation of premail. This information should not be necessary for ordinary use.

Multiple recipients

One of the tricky problems with mail encryption packages such as premail is how to deal with multiple recipients. Based on experience with previous versions, this version of premail tries very hard to "get it right." However, as a consequence, the exact behavior can sometimes be difficult to understand.

The hard part is when some of the recipients have encryption specified and others don't. What premail does is to split the recipients up into groups. If two recipients can receive the same actual message, they are in the same group, otherwise not. For example, recipients getting an encrypted and an unencrypted message cannot be in the same group. However, multiple recipients appearing in To: and Cc: fields that use the same encryption method will be in the same group. A single message, encrypted to multiple recipients, will be sent, which is considerably more efficient than encrypting separately for each recipient.

One subtle point is the handling of Bcc: recipients. The semantics of Bcc: specify that the mail be sent to each of the Bcc: recipients, but that none of the other recipients be able to find out their identity. However, encrypting to multiple recipients would defeat this, because it is possible to indentify all of the recipients of the encrypted message. Thus, each encrypted Bcc: recipient gets its own group.

Each recipient of an anonymous message also gets its own group, for similar reasons.

An attempt is made to make the headers in the message received by the recipient be the same as if no encryption were used. Specifically, the complete To: and Cc: header fields will be present, but the Bcc: field will be missing. One exception to this rule is anonymous messages, in which case the recipient can't see any information about the other recipients.

Error handling

The goal is to handle errors in the same way as sendmail. Thus, the exact handling depends on the setting of the -oe command line option. The default (as in sendmail) is -oep, meaning that the error message is printed to standard out, and the mail message is appended to the dead letter file (the location of which is a configuration option).

Another choice is -oem, in which case the error message and the mail message are packaged together and mailed back to the user. This is appropriate when the mailer has no way to deal with error messages returned from premail.

One additional choice, not provided by sendmail, is -oed, which prints the error message on standard out, but drops the mail message. This is a good choice if the mailer can interpret a non-zero return status code as indication of an error. This is the mode used by Netscape (and is automatically selected when premail is invoked as prezilla).

Security issues

In designing premail, usefulness and convenience were considered more important than top security. Nonetheless, it can provide good security, especially if you are aware of the security issues.

One overriding assumption was that your machine is secure, and that the serious threats were those of eavesdroppers on the network and e-mail forgers. In general, premail handles passive attacks quite well, while containing a number of vulnerabilities to active attacks.

Here are some potential security pitfalls with premail:

• Stores secrets information on disk file.
• Stores (potentially sensitive) temporary files on disk.
• Does not check authenticity of remailer list, remailer public key ring, or Mixmaster information gotten from the Web.
• Accessing the Web signals when anonymous mail is about to be sent, perhaps aiding traffic analysis.
• Does not evaluate the trustworthiness of public keys used for encryption and signature checking.

Useless features

Over the years, premail has accumulated a number of features of dubious value. One of them is support for MOSS, a nice encryption protocol that nevertheless failed to catch on. If you feel the urge to use it, documentation is available in the release notes for version 0.43.

One potentially cool feature is a server for decoding e-mail. This would be a useful feature if there were any mailers which used it. The protcol for the server was designed to be fast (much, much faster than invoking premail -decode separately for each message), as well as "crypto-neutral," meaning that it doesn't contain any features designed just for crypto, and that it could be used for other tasks, for example converting image formats or character sets. Thus, a client designed to use this protocol would like be fully exportable from the US. If you're interested in integrating support for this protocol into a popular e-mail client, please get in touch with me.

• The README file for premail version 0.33a.
• Release notes for version 0.43 of premail.