In that case, you might try what I said earlier. Log some unique test
string, then tail the last X number of lines of syslog and check for a
matching string. If that's all you can do, it's all you can do.
-Matt
On Mon, Sep 22, 2008 at 9:58 AM, Thomas Gelf <[EMAIL PROTECTED]> wrote:
> Hmmm... it's useless. I finally took the time to make some tests, it
> seems to be "impossible" to make PHP's syslog-calls fail. I changed
> permissions on /dev/log, I removed the device node - openlog() and
> syslog() always give back (boolean)true.
>
> Curious I also tried it from command line by calling /usr/bin/logger,
> the result was the same. Even if logger fails it gives me no error,
> see this strace snipplet as an example:
>
> > # strace logger test && echo "OK"
> > ...
> > connect(1, {sa_family=AF_FILE, path="/dev/log"}, 16) = -1 ENOENT
> > (No such file or directory)
> > close(1) = 0
> > exit_group(0) = ?
> > Process 4733 detached
> > OK
>
> So what would you suggest? Remove my tests for syslog() and openlog()
> return values from my Zend_Log_Writer_Syslog proposal? Or just leave
> them as they are? Can someone test behaviour on Windows 2003 R2? Are
> PHP's syslog functions aware of failing syslog calls on those systems?
>
> Kind regards,
> Thomas Gelf
>
> NB: I did all tests on Debian Etch and Debian Lenny, 64bit systems.
>
>
> Matthew Ratzloff schrieb:
>
>> Manually change the permissions for syslog on your local machine, make the
>> test fail, change the permissions to be correct and make sure the test
>> succeeds. Maybe log something then strpos the contents of tail. That's all
>> you can really do.
>>
>> -Matt
>>
> >
>
>> > I could achieve this by removing or changing permissions on /dev/log,
>> > no idea about Windows. But I've some doubt if doing so would make
>> > people running the tests happy :p
>> >
>> > Regards,
>> > Thomas
>>
>
>