On 6 December 2010 20:47, Steve Staples <sstaples@xxxxxxxx> wrote:
> On Mon, 2010-12-06 at 20:29 +0530, Ferdi wrote:
> > Greetings List members,
> >
> > I have a script that takes quite a while to run, one or two hours, I wish
> to
> > redirect the normal php output to a file on the webserver itself. I don't
> > mind if in the process, the browser displays a blank page. The reason I
> want
> > to do this is that if the script crashes or the browser Is closed by
> > mistake, I have absolutely no record of where the script stopped running.
> >
> > I could use code like below
> > At the beginning of the script:
> > ob_start();
> >
> > At the end of the script:
> > $page = ob_get_contents();
> > ob_end_flush();
> > $fp = fopen("output.html","w");
> > fwrite($fp,$page);
> > fclose($fp);
> >
> > However, I see some problems with this:
> > I'm not too sure of the size of the output. It may balloon to over the
> > buffering limit (in PHP? Apache?) and then what happens?
> > Secondly, if the script crashes before the end, I won't get any output.
> > Finally, I am using a library in the script that outputs status and error
> > messages of its own. So, if I manually opened a file and used fwrite()
> > alongside echo for my messages, I would lose out on those messages.
> >
> > Anybody has any pointers on how you could send the output not only to a
> > browser, but also to a file on the webserver? If not, at least to a file?
> >
> > Thanks and regards,
> > Ferdi
>
> Just curious, but if it takes that long to run, why are you running it
> from a browser? why not run it from the commandline, that way you dont
> have to change your php.ini for the webserver (increasing the timeout,
> memory limits, etc etc... you can change those for the CLI only?
>
> 2 hours is a long time to "hope" that the browser doesn't close, or
> connectivity doesn't get interupted for even 1 microsecond...
>
> if the script has "breaks" in it, where it starts to do something else,
> you can put in an email to yourself, to say "hey, we're HERE now"
>
> but i would look into running it from the CLI over the webserver, you
> would be less likely to run into issues on something that takes that
> amount of time to run.
>
> If you needed the output to be displayed on a webpage, you can write the
> progress to a file, and then have a php webpage that reads the file, and
> using ajax or whatever, refresh the content.
>
> good luck in your script, and if you still run it from the browser, and
> need to output to a file, then i would continually be writing content to
> that file, every time you do soemthing, or start another part of the
> script so you know EXACTLY where you are, at all times...
>
> Steve
>
Hi Steve,
Thanks for the tips. To answer your queries, I don't mind using CLI. How do
I then ensure the messages, error or otherwise, output by the library I use,
show up in the file I'm outputting to? Please note that I only make calls to
the functions and object methods from this library and error or success
messages are probably echo'd by the code from the library.
I believe some context is in order. I am actually sending an email to a
large number of our customers (around 10,000), each with the customer's
name.
My script does output a simple success or failure message. The library
outputs more technical info. All I would like to know is which email got
sent and which one failed. As you point out, I could write the success /
failure messages to a file, but, I would also like to capture the messages
output by the library.
Thanks and regards,
Ferdi
