Suzanne — Fri, 23 Jan 2004 14:29:35 +0000

I know a number of ISPs will block huge lists of BCCs as well, or terminate them at 150 or some arbitrary number.

I'm also leery of BCCs and somewhat guilty of emphasizing dynamic content so I don't have to deal with them...

Abhishek Reddy — Fri, 23 Jan 2004 09:59:23 +0000

m3rajk, my math is indeed flawed, as I noted in one of my responses to Suzanne above:

Quote: I said:
Suzanne, I was referring to Renegade's flawed code, which appends an address to the $sendto string, then sends out a set of mails, then appends the next address to $sendto, and then mails once again to the new set. I oversimplified the math so as not to break into this algorithm... it actually sends 200 emails to the first person, 199 to the second, 198 to the third . . . 1 to the last.

I don't know how else to send in the case of a dynamic body. Is there no way to construct an array of messages in PHP and send it all to the MTA once and have it mail each to a different address?

How are mails sent to really large lists of addresses in "industrial" situations? I guess some kind of third party software is used?

Mark Hensler — Fri, 23 Jan 2004 07:25:05 +0000

If the body of the email is different for each recipient, then you need one mail() call per recipient.

m3rajk — Thu, 22 Jan 2004 18:21:15 +0000

Abhishek Reddy wrote: As I understand it...

Suppose there are 200 addresses in the db to send to, the looping code with the stringed addresses will have the server send 200 mails out, 200 times (because the loop repeats it for each address in the db). 200*200 = 40,000. And since it sends to the same addresses, 200 people will theoretically end up with 200 copies of the same email in their inbox.

Why the server breaks is because the PHP mail() function is reasonably inefficient. For the reason that it opens a connection to the Mail Transfer Agent, sends the required mail data to the MTA, then closes the connection. This is fine for sending a few emails, or even for sending a list of emails for the MTA to parse and handle. But with the code in question, it means it will open a connection, send 200 addresses to send to and close the connection (this is fine), but repeat it 200 times (this is not fine).

As noted in the user comments on the ref.mail and function.mail pages on php.net, opening a single socket with the MTA to send multiple mails out is more efficient.

either i'm reading the code wrong, or your math is wrong.

i'm reading that in the one with the mail in the loop it adds an address with each loop, thus if there is 200 name,s the first gets 200 emails, the second 199, and thus down to the 200th gets one.

however i also see the point of what suzanne said about the dynamic body, in which case merely changing the script to reset the to feild should make it 200 emails flat int hat situation. and in that cas, do you have a way to send 200 different emails in one call? or would one need to make all 200 calls to the mail function?

personally if everything else was static, i'd build the mailing list as a bcc and send it once with myself in the to feild so that no recipient would see the others, thus protecting privacy

Mark Hensler — Thu, 22 Jan 2004 17:36:20 +0000

Ok, so maybe I didn't deep enough for some of you.

From what I've read: yes, yes, and "that can't be true!" (disbelief)

I didn't read the entire RFC, but I think Appendix B is describing how to handle TO, CC, and BCC headers.

   It is recommended that the UA provide its initial ("submission
   client") MTA with an envelope separate from the message itself.
   However, if the envelope is not supplied, SMTP commands SHOULD be
   generated as follows:

   1. Each recipient address from a TO, CC, or BCC header field SHOULD
      be copied to a RCPT command (generating multiple message copies if
      that is required for queuing or delivery).  This includes any
      addresses listed in a RFC 822 "group".  Any BCC fields SHOULD then
      be removed from the headers.  Once this process is completed, the
      remaining headers SHOULD be checked to verify that at least one
      To:, Cc:, or Bcc: header remains.  If none do, then a bcc: header
      with no additional information SHOULD be inserted as specified in
      [32].

'I have the impression that after the MTA recieves a full envelope, it scans the email headers in the DATA section for CC and BCC recipients. Then sends copies of the envelope out to each CC and BCC address (leaving out the BCC headers).

Abhishek Reddy — Wed, 21 Jan 2004 08:32:15 +0000

Mark, I didn't get a word of that.

Is this correct?
- MAIL command initiates a transaction
- RCPT does what? Identify a single primary recipient?
- DATA contains headers which are sent to the RCPT recipient? So one recipient gets all the headers?

At what point are the CC/BCC headers actually processed?

Mark Hensler — Wed, 21 Jan 2004 07:28:56 +0000

I've always avoided use of the BCC header in mail scripts. This is because of several (dated) documents that I have read that suggest that the BCC header is not fully and uniformally supported. I suppose the determining factor would be the MTA that you use to initially send out the email, but I have not done any in-depth investigation.

I also recently read that the To, Cc, and Bcc headers have nothing to do with determining where an email is sent to. But that the destination si determined in the SMTP negotiation through the RCPT command.

Imagine a 15 minute intermission while I run to the RFC Editor

Ah ha! I found it in Seciont 3.3 "Mail Transactions" of RFC 2821. Basically, there is an "envelope" which includes a MAIL command ("tells the SMTP-receiver that a new mail transaction is starting"), a RCPT command ("a forward-path identifying one recipient"), then a DATA command. The DATA command contains all information which will be sent to the address identified in the RCPT call. This data also includes the headers of the email (to, cc, bcc, etc.).

Quote: When RFC 822 format [7, 32] is being used, the mail data include the
memo header items such as Date, Subject, To, Cc, From.

I may have dug deaper than necessary. If your scripts work, pay no attention to my bcc phobia.

Suzanne — Wed, 21 Jan 2004 06:31:55 +0000

*understanding dawns*

Thank you, Abhi! I feel much less confused. It's a bad day when you think you know something, then start doubting yourself. Two Mondays in a row. *sigh*

Abhishek Reddy — Wed, 21 Jan 2004 05:46:15 +0000

Suzanne, I was referring to Renegade's flawed code, which appends an address to the $sendto string, then sends out a set of mails, then appends the next address to $sendto, and then mails once again to the new set. I oversimplified the math so as not to break into this algorithm... it actually sends 200 emails to the first person, 199 to the second, 198 to the third . . . 1 to the last.

As for efficiency, all reports I've read regarding this indicate that opening a single socket with the MTA (usually sendmail) is more efficient than using mail() repeatedly. I don't have first-hand experience with this so I can't say for myself. I objected to using a repeating mail() script in Renegade's loop since that would send out additional unnecessary mails... which could be dangerous.

Last comment on this page: http://nz2.php.net/manual/en/function.mail.php (lit at matches dot com[b])[/b] as well as the script you quoted, Suzanne.

druagord — Tue, 20 Jan 2004 21:26:45 +0000

IMO using BCC is better because it recipients won't be able to get the email adresse of other recipients spammer register themself to mailling list just to get emails. Server wise it make no difference.