Posted by James Taylor on 02/10/05 11:36

<snip>

> -bash-2.05b$ php test.php
> Content-type: text/html
> X-Powered-By: PHP/4.3.10
>
> Memory Limit: 32K
> Memory Used: 13432
> Error (1024) in /www/l-i-e.com/web/test.php:12 - Test
> ....-bash-2.05b$
>
> Note a complete lack of "Made it" nor an error message trapped by my error
> handler when the RAM runs out. :-(
>

This is the symptom I had when I recoded t osomething very similar.

> Looks like the Memory limit is hard-coded and won't trigger an error
> handler nor will it give you any chance to catch it.

Yes, This might be a mis-discription of error, that this error really is
php-fatal not just script-fatal.

<snip>

>>With error handling as normal, can it handle a fatal error? My issue is
>>not it would not report a fatal error, it does not continue with the
>>script.
>
>
> I think maybe not -- but maybe only specific errors exhibit this feature.
>
> Memory limit and Time limit could be "weird" enough to need it.
>
> You could try some other fatal errors to see.
>

Some errors do seem to be caught correctly such as Parse Error -
(especially with Eval) which highlights my theory of fatal being either
script-fatal or program-fatal. if its the PHP application that has died,
then it would explain why the memory limit and some others might not
execute correctly. I did not try time, but suspect that would be
script-fatal. Both these limits are designed to protect your computer
from crashing becuase of bad programmers (/me looks guilty), and it
would be daft to allow a programer to continue with code once either of
these have been breached.

>>The last issue I am worried about is scope - if I have run out of
>>memory, then what can I do - I seem to have issues running commands
>>which go over this limit during the command (irrelevant of their end
>>memory usage). Supposing I had enough memory to run the set_ini function
>>and increase my memory limit, then I would be able to execute any
>>command to roll back (transaction wise) the program execution so far,
>>and email/notify/whatever. This is irellevant if a "success" orientated
>>monitoring method is implemented.
>
>
> I think you would be well-served to figure out where the RAM is going, and
> making sure it's what it should be at various junctures.
>

yes, I seem to have a memory leak, which has been explained to me by a
collegue as a possibility that I am including a new object line inside
the loop (being used to Java's garbidge collector (and my manual calling
of it)) that would get recreated each loop, without much impact on
memory, and that removing the new call to outside the loop, and placing
a "refresh" or "renew" function into my object would save me memory.

This is a little concern as this is difficult to find in the code - the
only "large" (proportional to avail-mem) objects that Im using are in
loops of ten to twenty seconds (due to database calls).

> Maybe you're trying to patch symptoms instead of debugging the true problem.
>
> Just a thought.

Of course, an efficient elegant working program is always to be desired,
but the nature of this problem is potentially a PEBCAK - I can not
always limit the input ranges suitably (due to number of combinations of
different variables), perhaps I should write a calculate function which
takes the same arguments as the real function, and guesses to its length
of time to run and its memory usage. That way I can make a descision as
to whether to run it or not (of course it would take me a while to get
the memory profile accurate, but would be nice to see). This isnt as
easy as just saying if $x < 10 and $y < 10 or max $x must be inversly
proportional to $y (so as $y increases the max $x can be must decrease)
as it would involve actual data in the database and knowing how
different data affects the speed / memory. But thats my problem and very
specific to whats going on, so feel free to ignore...

> You could also maybe file this as a bug, after searching the bugs database
> to see if it's a known issue, or even a "feature"
>

I dont think it is a bug I think it makes sense in its repeats, but I
think I'll add a comment to the page on the php manual with my findings,
and link to this discussion.

Thanks for that

[Back to original message]