Bodecs Bela (2018-03-21):
> Should not the unix_write behaves the same? so, when
>
> ret = send(s->fd, buf, size, MSG_NOSIGNAL);
>
> return 0 in case of SOCK_STREAM type, then the unix_write() should return
> AVERROR_EOF instead of 0?
No: there is no EOF reporting on writing, since the
2018.03.20. 22:41 keltezéssel, Nicolas George írta:
Bodecs Bela (2018-03-20):
start a. and in a separate terminal window start b. After 10 seconds a.
terminates but b. remains running on infinite time.
Ok, got it.
But I expected b. to
2018.03.20. 22:41 keltezéssel, Nicolas George írta:
Bodecs Bela (2018-03-20):
start a. and in a separate terminal window start b. After 10 seconds a.
terminates but b. remains running on infinite time.
Ok, got it.
But I expected b. to
Bodecs Bela (2018-03-20):
> start a. and in a separate terminal window start b. After 10 seconds a.
> terminates but b. remains running on infinite time.
Ok, got it.
> But I expected b. to
> terminate after 1 sec (100 microsec) when no
2018.03.20. 21:23 keltezéssel, Nicolas George írta:
Bodecs Bela (2018-03-20):
I am sorry, but you misunderstood me. When recv() return 0 it means that no
data was read.
And for the third time I am asking how you observe that. What is your
testing procedure?
Regards,
ah, sorry. Let's see
Bodecs Bela (2018-03-20):
> I am sorry, but you misunderstood me. When recv() return 0 it means that no
> data was read.
And for the third time I am asking how you observe that. What is your
testing procedure?
Regards,
--
Nicolas George
signature.asc
Description: Digital signature
2018.03.20. 20:58 keltezéssel, Nicolas George írta:
Bodecs Bela (2018-03-20):
see the retry_transfer_wrapper() in avio.c If
ret = transfer_func(h, buf + len, size - len);
is always zero, so it gets into infinite loop.
transfer_func is unix_read in case of unix protocol.
Let me be more
2018.03.20. 20:56 keltezéssel, Bodecs Bela írta:
2018.03.20. 20:50 keltezéssel, Nicolas George írta:
Bodecs Bela (2018-03-20):
avio rw_timeout handling in retry_transfer_wrapper() is based on
returning
EAGAIN from protocols' read function. unix_read function returns 0 in
case of no data
Bodecs Bela (2018-03-20):
> see the retry_transfer_wrapper() in avio.c If
>
> ret = transfer_func(h, buf + len, size - len);
>
> is always zero, so it gets into infinite loop.
>
> transfer_func is unix_read in case of unix protocol.
Let me be more accurate: I am pretty sure you analysis is
2018.03.20. 20:50 keltezéssel, Nicolas George írta:
Bodecs Bela (2018-03-20):
avio rw_timeout handling in retry_transfer_wrapper() is based on returning
EAGAIN from protocols' read function. unix_read function returns 0 in
case of no data was read. It happens even if timeout it set for a
Bodecs Bela (2018-03-20):
> avio rw_timeout handling in retry_transfer_wrapper() is based on returning
> EAGAIN from protocols' read function. unix_read function returns 0 in
> case of no data was read. It happens even if timeout it set for a valid
> value and thus rw_timeout handling can not work
Dear All,
avio rw_timeout handling in retry_transfer_wrapper() is based on returning
EAGAIN from protocols' read function. unix_read function returns 0 in
case of no data was read. It happens even if timeout it set for a valid
value and thus rw_timeout handling can not work and wait for ever.
12 matches
Mail list logo
