On 2014-09-23 10:47, Michael Wallner wrote:
> On 2014-09-23 10:11, Sanford Whiteman wrote:
>>> if somehost.example.com has the MX, it should return true with
>>> checkdnsrr('somehost.example.com'). If example has the MX set to
>>> somehost.example.com or similar, it should return true as well. Or am
>>> I missing your point?
>>
>> You are missing it, as there are no MX records involved. I'm
>> demonstrating that the function gives incorrect results in the absence
>> of an MX record because there is a bug in CNAME handling (or a bug in
>> A handling, if you prefer).
>>
>> somehost.example.com has no MX RR. It has a CNAME (per DNS rules, the
>> CNAME must thus be the only record for somehost.example.com). The
>> CNAME points to example.com.
>>
>> example.com has no MX RR. It has an A pointing to 1.2.3.4.
>>
>> checkdnsrr('somehost.example.com') should return false. It returns true.
>> checkdnsrr('example.com') should return false. It returns false.
>>
>> You could try to revise the docs and say, "Oh, it doesn't really mean
>> an MX lookup, it means an explicit MX *or* implicit MX, like how a
>> basic smtp-sender works. So that's why the CNAME --> A one returns
>> true." Except then 'example.com' should return true as well. It does
>> not: it returns false. Changing the docs does not fix the bug.
>
> Ah okay, I think you're right. It shouldn't accept the CNAME.
>
Looking at the source, though, neither the code nor the docs make any
assertions on the actual type of response they got. So it could still be
seen as "works as expected", because it just checks if any answer is
received. If that functionality is useful could be debatable.
--
Regards,
Mike
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
