Hi,
I have tried you patches, it worked only once out of five trials...
Maybe the tcflush() function never works with my IrDA dongle?
nsc-ircc, Found dongle: HP HSDL-1100/HSDL-2100
[EMAIL PROTECTED]:/tmp$ scmxx -i -vvv
Accessing device /dev/ircomm0...done
Using "UTF-8" as system character set.
Sending command: ATZ
Received: +++
Value matched nothing.
Value matched nothing.
Received: ERROR
Value matched ERROR.
Value matched ERROR.
Error: cannot communicate with device.
[EMAIL PROTECTED]:/tmp$ scmxx -i -vvv
Accessing device /dev/ircomm0...done
Using "UTF-8" as system character set.
Sending command: ATZ
Received: +++
Value matched nothing.
Value matched nothing.
Received: ERROR
Value matched ERROR.
Value matched ERROR.
Error: cannot communicate with device.
[EMAIL PROTECTED]:/tmp$ scmxx -i -vvv
Accessing device /dev/ircomm0...done
Using "UTF-8" as system character set.
Sending command: ATZ
Received: ERROR
Value matched ERROR.
Value matched ERROR.
Received: OK
Value matched OK.
Value matched OK.
OK, a modem device is present.
[.........]
[EMAIL PROTECTED]:/tmp$ scmxx -i -vvv
Accessing device /dev/ircomm0...done
Using "UTF-8" as system character set.
Sending command: ATZ
Received: +++
Value matched nothing.
Value matched nothing.
Received: ERROR
Value matched ERROR.
Value matched ERROR.
Error: cannot communicate with device.
[EMAIL PROTECTED]:/tmp$ scmxx -i --start-delay=1 -vvv
Accessing device /dev/ircomm0...done
Waiting for 1 seconds as requested...
Using "UTF-8" as system character set.
Sending command: ATZ
Received: +++
Value matched nothing.
Value matched nothing.
Received: ERROR
Value matched ERROR.
Value matched ERROR.
Error: cannot communicate with device.
[EMAIL PROTECTED]:/tmp$ scmxx -i --start-delay=5 -vvv
Accessing device /dev/ircomm0...done
Waiting for 5 seconds as requested...
Using "UTF-8" as system character set.
Sending command: ATZ
Received: ERROR
Value matched ERROR.
Value matched ERROR.
Received: Z
Value matched nothing.
Value matched nothing.
Error: cannot communicate with device.
On 7/2/05, Hendrik Sattler <[EMAIL PROTECTED]> wrote:
> Am Freitag, 1. Juli 2005 17:39 schrieb SDiZ:
> > I guess there are some old data in the read buffer, because of what i
> > see in the debug output:
> > Sending command: ATZ
> > Received: ERROR
> > Value matched ERROR.
> > Value matched ERROR.
> > Received: CPM=
> > Value matched nothing.
> > Value matched nothing.
> > Received: OK
> > Value matched OK.
> > Value matched OK.
> > OK, a modem device is present.
>
> Then maybe the order is wrong. Currently (and with your patch), we open the
> device and flush the buffers. However, the switch to RCCP mode is done after
> flushing the buffers.
> The patch for my suggestion looks like this:
> --- actions.c 8 Jun 2005 17:49:17 -0000 1.30
> +++ actions.c 1 Jul 2005 15:57:13 -0000
> @@ -108,6 +108,10 @@
> int retval = 1;
> char* ack;
>
> + /* asume that we previously waited the 1000ms */
> + tty_write("+++",3);
> + usleep(1010*1000);
> + tty_flush();
> tty_write("\r",1);
> at_command_send(AT_GEN_INIT,NULL);
> ack = at_read_line();
> --- ttyaccess.c 8 Jun 2005 17:49:18 -0000 1.25
> +++ ttyaccess.c 1 Jul 2005 15:57:13 -0000
> @@ -69,6 +69,10 @@
> }
> #endif
>
> +void tty_flush () {
> + tcflush(tty_portfd, TCIOFLUSH); //clear serial tty
> +}
> +
> #define NONBLOCK_OPEN 1
> void tty_open(struct port_args_t* args) {
> #ifdef OS2
> @@ -195,7 +199,7 @@
> newtio.c_cc[VMIN]=0; //return after timeout, even with nothing
> }
> newtio.c_cc[VTIME]=args->timeout; //try reading for this amount of time (in
> deciseconds)
> - tcflush(tty_portfd, TCIOFLUSH); //clear serial tty
> + tty_flush();
> if(tcsetattr(tty_portfd,TCSANOW,&newtio) < 0){ //set now
> fprintf(stderr,"%s: %s\n",_("Error in setting port
> attributes"),strerror(errno));
> exit(EXIT_FAILURE);
> --- ttyaccess.h 10 Jun 2005 10:08:40 -0000 1.6
> +++ ttyaccess.h 1 Jul 2005 15:57:13 -0000
> @@ -45,6 +45,7 @@
>
> int tty_write (const char* data, size_t count);
> char* tty_read_line (int (*stop_condition)(const char*,const size_t));
> +void tty_flush ();
>
> // \x1a represents Ctrl-Z (represents end of data stream)
> #define tty_write_data(data,count) (tty_write(data,count) &&
> tty_write("\x1a",1))
>
>
> I'm somewhat convinced that there is no need to send '\r' twice.
> The "+++" should tell the phone to stop sending _immediately_ but we all know
> buggy things ;)
>
> HS
>
