Create "please wait" page while process running

I had thought of refreshing the page, but I have many variables which are passed as _POST vars including full text and html versions of the message which could get quite lengthy for the subsequent _GET post you suggest above. What is the limit for a QUERY_STRING?

you mean max url size? i think 1024, but i'm not 100% sure.

You mailing system is quite poor, you cannot rely on browser's request for such activity, because any interruption of the request (browser timeout, network problems, and so on) would stop the process.
I would rather use a crontab entry that triggers a php script (using the CLI php binary) that every 5 minutes polls file X for its existence. If it exists, opens it for reading and fetches your POST variables previously saved and then starts sending mail. In this case there is no time limit.

Thanks CrYpTiC_MauleR

I think you are on the right track. I have rebuilt me page to process 100 at a time and then I use header(Location: ...) to process the next group. This works well and as I need. I was looking at your suggestion above with the Refresh which would let me print to that window in addition to reloading it. I would like to do that, but am concerned that if it takes more than 5 seconds to process a batch, then the page will be refreshed calling the next group and the tail of the previous group will get cut off.

Does this make sense? If I use "ignore_user_abort" will the previous batch continue even after the reload in 5 seconds?

Keep in mind an error on the client would cause screwups in the sending, i.e. the browser crashes (for whatever reason) and messages have are sent out to indivduals twice or more.

IF you plan to go ahead with the process-100-per-page plan, I recommend you put in handling for the event the client computer starts a mail batch, but never completes it - allow them to resume from the last point in the batch.

Otherwise, for the easier option of just doing it all in one go it should be quite easy to send a period every 10 messages.

Simple example:
[code]
ignore_user_abort(true); // keep processing even if the client goes away
echo "started mail batch, each "." means 10 messages have been sent<br />";
$sql = "select user_email, user_firstname, user_lastname FROM user_table WHERE user_in_mailinglist = 1";
$query = mysql_query($sql) or die("mysql error: ".mysql_errnum().":- ".mysql_error()."<br />".$sql);
$rows = mysql_fetch_array($query);
for($loopy = 0; $loopy < count($rows); $loopy++){
   /*
    * Mail stuff here
    * mail($rows[$loopy][0], $email_subject, $email_body);
    */
   if(($loopy %10) === 0)
      echo ".";
}
if(connection_aborted()){

}
echo "mail batch complete!<br />";
[/code]

... damn tab button... damn space bar. Damn inability to use a computer.

I didn't quite finish that:

[code]
ignore_user_abort(true); // keep processing even if the client goes away
/*
 * Make a log that the mail batch began
 * $sql = "INSERT INTO mail_batches VALUES ( DATE(), 'Begun processing the batch (started by '.$_SERVER['REMOTE_ADDR'].')')";
 * query($sql);
 */
echo "started mail batch, each "." means 10 messages have been sent<br />";
$sql = "select user_email, user_firstname, user_lastname FROM user_table WHERE user_in_mailinglist = 1";
$query = mysql_query($sql) or die("mysql error: ".mysql_errnum().":- ".mysql_error()."<br />".$sql);
$rows = mysql_fetch_array($query);
for($loopy = 0; $loopy < count($rows); $loopy++){
   /*
    * Mail stuff here
    * mail($rows[$loopy][0], $email_subject, $email_body);
    */
   if(($loopy %10) === 0){
      echo ".";
   }
   if(connection_aborted()){
      /*
       * Make a log that the mail batch completed
       * $sql = "INSERT INTO mail_batches VALUES ( DATE(), $_SERVER['REMOTE_ADDR'].' stopped watching the batch process')";
       * mysql_query($sql);
       */
   }


}
/*
 * Make a log that the mail batch completed
 * $sql = "INSERT INTO mail_batches VALUES ( DATE(), 'Completed processing the batch (started by '.$_SERVER['REMOTE_ADDR'].')')";
 * mysql_query($sql);
 */
echo "mail batch complete!<br />";
[/code]

The solution above seems fine as long as you can ensure that you have control over the php script timeout - if you don't and it's set to a low value then it will still quit.  I've done something remotely similar which I've used to avoid a script timeout on a photo website for resizing images.  You need to write the page so that each time through the script it can get a list of e-mail addresses still due to be sent - easiest way would be to store all the data for sending the messages in the session, with the destination addresses in a queue or similar.

Next thing is to set the META refresh tag to reload the page after the same time as your script timeout.  This should mean that it works on simple browsers.

In your php script, you need to record the time the script started running, and the number of messages processed so far in this execution.  At the start of each time through the loop, use this information to predict whether the timeout will occur this time round.  If it does, don't execute and echo a javascript command to refresh the page.  Then stop processing and wait.  If the browser supports Javascript it will immediately refresh and carry on.  If JavaScript is disabled, the meta tag should cause the page to be refreshed.

If you are interested in this approach, I can dig out sample code.  I find it very effective for running long repeated jobs on machines where the script timeout is set too low.  It should also get around your problem of the browser timing out, and you can add progress information to the page each time it is reloaded if you wish.

I thought I had a great solution with the header(Location) to the script calling the next 100 addresses, but ran into this problem. If I had too many addresses, my browser gave me a message that there were too many redirects and it was giving up. Bummer.

I suspect using header(Redirect: 5; Location...) would do the same.

I hate to take this alternative, but I might be forced to create a file and have a cron job run every few minutes.

Suggestions?

"You may also try to spawn a background process yourself.. "

Hmmmmm, this sounds familiar. I wonder where I heard this before?

Alan

> Hmmmmm, this sounds familiar. I wonder where I heard this before?

You are right AlanJDM, I didn't notice your post, i'm sorry. Please ignore my previous comment.

I agree that the spawning technique would appear to be the best.  I wrote mine to get around safe mode on a webserver, where there is no chance I can spawn external processes.  If you can use something in the background, it would be much better.

Hey AlanDJM, Can you explain how you do this...

>At the end of that script I set a flag in a table that indicate it is done running. That way all my "please wait" page does >is refresh every 10 seconds checking that flag. This way there is no timing issues and worrying about refreshing before >it is safe to do so.

How does a newly refreshed page check to see if a speerate process is completed?

Thom

The script that runs in background will update a field in a table to a value, I would use 1 and 0. At the beggining of the script it would set this value to 1 which will indicate that the script is running, and at the end of the script it will change the value to 0 which indicates it is not running.

Then, the page that gets refreshed will contain a script that queries that value. It will continue to refresh until it sees a 0. When it sees a 0 it will know the script is done running and react accordingly, probably with a header call back to the original page with a message that says "complete".


Alan

Alan, nice idea. I guess the only problem could be too many redirects (from waiting too long) causing the borwser to stop trying to load the page like I got when I tried to reload every 10 emails.

No, there is only one redirect... the end of the script. The only script that will run a long time is the script in background. The script that is refreshing will only redirect once, after the flag gets set. There is nothing to timeout.


Alan