Daniel,

On Fri May 13 2005 at 22:02, Daniel Stenberg <daniel-curl_at_haxx.se> wrote:

>> According to "curl -v <url>", it gets all the page content (up to the
>> end), but then waits for "something". When built under 10.3 or even
>> under 10.4 with HAVE_FINE_POLL (manually) undefined, it downloads the
>> page content and "closes" the connection normally.
>
> I guess this calls for me to adjust the test. Unfortunately, it would
> have to be a check for OS and then always disable poll on Mac OS X,
> since I don't know exactly what features in poll() that doesn't work
> in 10.4. Obviously the waiting-on-nothing check that used to not work
> isn't good enough to detect a bad poll().

Well, just in case, I'm not sure that that's bad "poll" that causes
the malfunction - it just happens that switching back to old schema
solves the problem.. But you know it better. ;)

And, I guess, you know that that configure test under 10.3 fails
because of the *assertion* in system poll on NULL argument (I'm not
sure if the assertion is the expected result).

>> The strange thing is that Mac OS X "built-in" curl (7.13.1) works
>> fine. According to curl-config it's indistinguishable from the
>> version that I build manually, but apparently it is..
>
> Have you tried to strace (or similar, I believe it might be called
> ktrace) it to see if it differs from your own build?

I just did it (with ktrace): yes, the built-in version doesn't use
poll, but uses select.

Regards,
Grigory
Received on 2005-05-13