How can I submit standalone programs with php?

Debbie-Leigh

Hi,

I'm building an application which submits scheduled php jobs. The way I would like it to work is to have one php program to submit the other php programs on a scheduled basis.

This control program shouldn't have to wait until one job finishes before it can start the next one; instead, it should be able to submit all the jobs in one go, but it also needs to be able to be informed of the return codes of the submitted programs so dependent programs can be submitted too.

I've looked at the exec() and system() commands, but they either don't seem to allow return codes to be returned from a submitted program to the submitter program or they unconditionally print anything the submitted programs produce.

So, does anyone know of any techniques to use these or other commands to achieve what I'm after?

Debbie

Drakla

Exec does allow you to find out what the return value is as the third parameter

string [man]exec[/man] ( string command [, array &output [, int &return_var]] )

System lets you get the returned value as the 2nd parameter

string [man]system[/man] ( string command [, int &return_var] )

And to get rid of any output just redirect command output to the deadzone

php yourscript.php > /dev/null 2>&1 &

On windows just direct it into a file you can nuke or rewrite over

LoganK

Drakla wrote:
On windows just direct it into a file you can nuke or rewrite over

echo "blah" > NUL

That's windows' null device

Debbie-Leigh

Just a couple of questions:

To get the return code, is it just the same as normal i.e. whatever is on the return statement is passed back to the calling program?
I gather this redirect command is for the command line. Where would I put it in code - on the calling statement or in the called program? Also, would it be the same statement or something different in code?

Debbie

Drakla

As the 3rd parameter is passed by reference you use it like this:

define ('NEWSLETTER_COMMAND', 'php newsletter-mailer.php > /dev/null 2>&1 &');


exec (NEWSLETTER_COMMAND, $out, $retCode);

echo "Your return value is: ", $retCode;

Doesn't have to be php you're executing of course, can be any shell command like

$command = 'unzip -o '.$skinzip.' '.SKIN_CUSTOM_DIR.'"*" -d '.SKIN_CUSTOM_DIR_PATH;

Actually I just remembered, if you redirect the output your script doesn't wait for it to return, that's why I do it in the newsletter, so that the script kicking it off doesn't wait about. The unzip command will wait, however, and give you a return code (just please don't be 11).

Debbie-Leigh

That seems quite straight forward, but does the command have to be assigned to a constant? Can it be put into a variable or directly into the exec command as well?

And could you explain what the 2>&1 & bit means?

Oh and just curious - what does a return code of 11 mean? :-)

Debbie

Drakla

No, you don't have to put it in a constant, I just have a config file and put it in that.

2>&1 means redirect everything sent to the error stream to the standard output stream, and the > /dev/null means redirect output to nowhere, so all script output is basically binned.

And with unzip the return code of 11 means I've messed things up again.

Debbie-Leigh

Oh, I see. Are these numbers the same as those the manual mentions for proc_open?

What does the last & do, though? Is it needed?

Drakla

Yes, same as proc. The last & is required if you want to 'background' what you're running. If you use that exec will move straight on after kicking the command off and not wait for it to return. In my mass mailer the script starts it going then tells you it's up and running, and an hour later the script emails me to tell me all went fine, but of course you don't get the return value from the script / command - so if you wanted background processing and return values the exec'd script would have to have some way of registering its result.