Looking for a more clever solution

[t***@aplawrence.com]

I have a client with an app that uses a modem to page technicians.
Doesn't want to do that; wants to sdend the data to their cell phone
instead.
Unfortunately, app apparently does more than just open the port
and send AT commands. Looks like it is actually doing ioctl stty's
and expecting happy results. App vendor (of course) is unhelpful
so far; wants customer to upgrade to Windows (which he'll happily
do when hell freezes solid, or so he says).

Tell the app to use a pty. Have a daemon listen to the master side for
messages.

Hmm. I never realized I could do that. I always thought you had to
open both, exec what you wanted to read on the slave, and read/write
from the master.

Searching Google for examples that don't follow that model has been
fruitless, and so far my experimentation with simple tests hasn't
let me read from ptyx anything written to ttyx.. can you give a
simple example?

Probably should move this to c.u.s.p


--
***@aplawrence.com Unix/Linux/Mac OS X resources: http://aplawrence.com
Get paid for writing about tech: http://aplawrence.com/publish.html

[John DuBois]

Post by t***@aplawrence.com

I have a client with an app that uses a modem to page technicians.
Doesn't want to do that; wants to sdend the data to their cell phone
instead.
Unfortunately, app apparently does more than just open the port
and send AT commands. Looks like it is actually doing ioctl stty's
and expecting happy results. App vendor (of course) is unhelpful
so far; wants customer to upgrade to Windows (which he'll happily
do when hell freezes solid, or so he says).

Tell the app to use a pty. Have a daemon listen to the master side for
messages.

Hmm. I never realized I could do that. I always thought you had to
open both, exec what you wanted to read on the slave, and read/write
from the master.

Nope. With earlier versions of the pty driver you had to open the master side
first; in the current version you can open either. The slave side can be
closed & reopened without having to reopen the master side. If the master side
is closed, the slave side will see a "hangup". This works for your purposes
because it's the master side you will want to keep constantly open, with
messages being read from it.

Post by t***@aplawrence.com
Searching Google for examples that don't follow that model has been
fruitless, and so far my experimentation with simple tests hasn't
let me read from ptyx anything written to ttyx.. can you give a
simple example?
Probably should move this to c.u.s.p

Find a pty that isn't in use (for your purposes, you will need to reserve a
fixed pty, perhaps by selecting a high-numbered pty and opening it early in rc2
processing). Then (using ttyp63 for an example):

On one screen, do:
$ cat /dev/ptyp63

From another screen, echo or cat to /dev/ttyp63. The above cat process should
spit out everything you send to the slave.

John

[t***@aplawrence.com]

Post by John DuBois

Post by t***@aplawrence.com

I have a client with an app that uses a modem to page technicians.
Doesn't want to do that; wants to sdend the data to their cell phone
instead.
Unfortunately, app apparently does more than just open the port
and send AT commands. Looks like it is actually doing ioctl stty's
and expecting happy results. App vendor (of course) is unhelpful
so far; wants customer to upgrade to Windows (which he'll happily
do when hell freezes solid, or so he says).

Tell the app to use a pty. Have a daemon listen to the master side for
messages.

Hmm. I never realized I could do that. I always thought you had to
open both, exec what you wanted to read on the slave, and read/write
from the master.

Nope. With earlier versions of the pty driver you had to open the master side
first; in the current version you can open either. The slave side can be
closed & reopened without having to reopen the master side. If the master side
is closed, the slave side will see a "hangup". This works for your purposes
because it's the master side you will want to keep constantly open, with
messages being read from it.

Post by t***@aplawrence.com
Searching Google for examples that don't follow that model has been
fruitless, and so far my experimentation with simple tests hasn't
let me read from ptyx anything written to ttyx.. can you give a
simple example?
Probably should move this to c.u.s.p

Find a pty that isn't in use (for your purposes, you will need to reserve a
fixed pty, perhaps by selecting a high-numbered pty and opening it early in rc2
$ cat /dev/ptyp63
From another screen, echo or cat to /dev/ttyp63. The above cat process should
spit out everything you send to the slave.

Hmm. I tried that on my Mac while I was away for the weekend, did not
work. I'm back home, firing up the SCO box..

Son of a gun. Wonder what is up with BSD that makes that different.
Oh well, this is for SCO, and that works fine. I'll look at the
BSD stuff another day.

Thanks John. That's such a simple thing, but I never knew it could be done!

--
***@aplawrence.com Unix/Linux/Mac OS X resources: http://aplawrence.com
Get paid for writing about tech: http://aplawrence.com/publish.html