TCP/IP

I’m having two problem with TCP/IP (either 4.25 or 5.0). These problem
do not show in QNX6 or Windows.

First problem is if I send() a block of 300K (apparently anything bigger
then 65K)
the receving end gets all the data but it’s corrupted,looks like it’s out of
order ;-(
All this with TCP not UDP.

Second problem which has already been discussed in other post:
Unless the connection is kept alive by sending data at least
every 300ms, the first send() will take a very long time compare to
all subsequent send(). TCP_NODELAY doesn’t improve the situation.

Again I want to point out these problems do not show in other OSes,
including QNX6.

  • Mario

Mario Charest postmaster@127.0.0.1 a écrit dans le message :
afcp4h$c13$1@inn.qnx.com

I’m having two problem with TCP/IP (either 4.25 or 5.0). These problem
do not show in QNX6 or Windows.

First problem is if I send() a block of 300K (apparently anything bigger
then 65K)
the receving end gets all the data but it’s corrupted,looks like it’s out
of
order ;-(
All this with TCP not UDP.

Second problem which has already been discussed in other post:
Unless the connection is kept alive by sending data at least
every 300ms, the first send() will take a very long time compare to
all subsequent send(). TCP_NODELAY doesn’t improve the situation.

Again I want to point out these problems do not show in other OSes,
including QNX6.

I see :wink:

  • Mario

    \

Actually, the second problem is also encontoured with RTOS 6.2 as i
experienced.

inn.qnx.compostmaster@127.0.0.1 wrote in message
news:afgm5b$8pt$1@inn.qnx.com

Mario Charest postmaster@127.0.0.1 a écrit dans le message :
afcp4h$c13$> 1@inn.qnx.com> …


I’m having two problem with TCP/IP (either 4.25 or 5.0). These problem
do not show in QNX6 or Windows.

First problem is if I send() a block of 300K (apparently anything bigger
then 65K)
the receving end gets all the data but it’s corrupted,looks like it’s
out
of
order ;-(
All this with TCP not UDP.

Second problem which has already been discussed in other post:
Unless the connection is kept alive by sending data at least
every 300ms, the first send() will take a very long time compare to
all subsequent send(). TCP_NODELAY doesn’t improve the situation.

Again I want to point out these problems do not show in other OSes,
including QNX6.

I see > :wink:

\

  • Mario



    \

Previously, Ron Cococcia wrote in qdn.public.qnx4:

Hi,
I’m writing a diagnostic program for our product, and one thing I’d like to
do is determine if the loader is QNX or not. What methods can I use to
determine if the QNX loader is being used?

TIA,
Ron

One way would be to read block 0 of the hard disk (/dev/hd0) and
scan it for the string QNX.

“Pascal Bouchard” <pasbou@fenclo.com> wrote in message
news:agib51$1nj$1@inn.qnx.com

Actually, the second problem is also encontoured with RTOS 6.2 as i
experienced.

Yep my mistake. With 6.2 delay is not there if using 127.0.0.1 but it’s
there if using ip address of interface. (running server and client on same
machine)

inn.qnx.compostmaster@127.0.0.1 wrote in message
news:afgm5b$8pt$> 1@inn.qnx.com> …

Mario Charest postmaster@127.0.0.1 a écrit dans le message :
afcp4h$c13$> 1@inn.qnx.com> …


I’m having two problem with TCP/IP (either 4.25 or 5.0). These
problem
do not show in QNX6 or Windows.

First problem is if I send() a block of 300K (apparently anything
bigger
then 65K)
the receving end gets all the data but it’s corrupted,looks like it’s
out
of
order ;-(
All this with TCP not UDP.

Second problem which has already been discussed in other post:
Unless the connection is kept alive by sending data at least
every 300ms, the first send() will take a very long time compare to
all subsequent send(). TCP_NODELAY doesn’t improve the situation.

Again I want to point out these problems do not show in other OSes,
including QNX6.

I see > :wink:

\

  • Mario





    \