Prev: New proxy site 100% free
Next: does pthread on Linux use clone()?
From: K-mart Cashier on 5 Jul 2010 17:25
Let's say I run the 'who' command on the z-shell. Like the following..

[cdalten(a)localhost ~]$ who
cdalten pts/1 2010-06-17 16:42 (:0.0)
cdalten pts/2 2010-06-13 12:49 (:0.0)
cdalten pts/3 2010-06-05 18:32 (:0.0)
cdalten pts/4 2010-06-18 17:49 (:0.0)
cdalten pts/5 2010-06-29 15:26 (:0.0)
cdalten pts/6 2010-07-05 14:21 (:0.0)
cdalten pts/8 2010-06-10 20:51 (:0.0)
[cdalten(a)localhost ~]$

From my understanding, the output from 'who' goes to the stdout. The
question is, how does this go to my terminal? Does the shell access
the stdout stream?


Chad
From: pk on 5 Jul 2010 17:31
On Mon, 5 Jul 2010 14:25:30 -0700 (PDT) K-mart Cashier <cdalten(a)gmail.com>
wrote:

> Let's say I run the 'who' command on the z-shell. Like the following..
>
> [cdalten(a)localhost ~]$ who
> cdalten pts/1 2010-06-17 16:42 (:0.0)
> cdalten pts/2 2010-06-13 12:49 (:0.0)
> cdalten pts/3 2010-06-05 18:32 (:0.0)
> cdalten pts/4 2010-06-18 17:49 (:0.0)
> cdalten pts/5 2010-06-29 15:26 (:0.0)
> cdalten pts/6 2010-07-05 14:21 (:0.0)
> cdalten pts/8 2010-06-10 20:51 (:0.0)
> [cdalten(a)localhost ~]$
>
> From my understanding, the output from 'who' goes to the stdout. The
> question is, how does this go to my terminal? Does the shell access
> the stdout stream?

stdout is file descriptor 1 in the shell, and the "who" process which is
a child of the shell inherits it.
To have the text appear on the terminal, file descriptor 1 must be
connected to a terminal device, so when something is written to file
descriptor 1, it is really written to the terminal.

Under most OSs, to see which terminal your shell is using, you can use
the "tty" command:

$ tty
/dev/pts/4

I suppose you're wondering what kind of device a terminal is. The accurate
answer is *very* complicated; for the purpose of your question, it can be
seen as a device driver in the OS kernel that knows how to read and
interpret your keystrokes and to write to the output device, be it a real
physical terminal (eg connected to a serial port) or a virtual one (for
example a console or an xterm).

So to recap (and simplifying): "who" writes something to file descriptor 1,
using write(); write() is implemented in the kernel, which sees that the
supplied descriptor 1 is connected to a terminal, and thus delegates the job
of writing to the terminal device driver, which knows how to do that (this
last action can itself involve quite a lot of other steps,
especially in a graphic environment).
From: Rainer Weikusat on 5 Jul 2010 17:58
K-mart Cashier <cdalten(a)gmail.com> writes:
> Let's say I run the 'who' command on the z-shell. Like the following..
>
> [cdalten(a)localhost ~]$ who
> cdalten pts/1 2010-06-17 16:42 (:0.0)
> cdalten pts/2 2010-06-13 12:49 (:0.0)
> cdalten pts/3 2010-06-05 18:32 (:0.0)
> cdalten pts/4 2010-06-18 17:49 (:0.0)
> cdalten pts/5 2010-06-29 15:26 (:0.0)
> cdalten pts/6 2010-07-05 14:21 (:0.0)
> cdalten pts/8 2010-06-10 20:51 (:0.0)
> [cdalten(a)localhost ~]$
>
> From my understanding, the output from 'who' goes to the stdout. The
> question is, how does this go to my terminal? Does the shell access
> the stdout stream?

The stdout stream is an abstract object provided by stdio (part of the
C-library). The basic idea behind that is that people who want to do
'input' and 'output' can deal with logical units of data (eg
'characters' or 'lines') while the library takes care of a sensible
buffering strategy for interaction with the OS. The 'stdout' file
descriptor is file descriptor #1 and the file description (in the
kernel) associated with that refers to the 'output channel' of your
'terminal'. Usually, this will be a pty slave, as in your output
above.
From: Tim Harig on 5 Jul 2010 18:09
On 2010-07-05, K-mart Cashier <cdalten(a)gmail.com> wrote:
> Let's say I run the 'who' command on the z-shell. Like the following..
>
> [cdalten(a)localhost ~]$ who
> cdalten pts/1 2010-06-17 16:42 (:0.0)
> cdalten pts/2 2010-06-13 12:49 (:0.0)
> cdalten pts/3 2010-06-05 18:32 (:0.0)
> cdalten pts/4 2010-06-18 17:49 (:0.0)
> cdalten pts/5 2010-06-29 15:26 (:0.0)
> cdalten pts/6 2010-07-05 14:21 (:0.0)
> cdalten pts/8 2010-06-10 20:51 (:0.0)
> [cdalten(a)localhost ~]$
>
> From my understanding, the output from 'who' goes to the stdout. The
> question is, how does this go to my terminal? Does the shell access
> the stdout stream?

The shell inherets its stdout from login. login inhereted stdout from
getty. Getty opened the file descriptor, which is probably a pipe to the
kernel's tty subsystem. The shell leaves its stdout open when it
fork()/exec()s and the file descriptor remains open after the exec(). who
may then write to stdout as it would to any other file descriptor.
From: Nicolas George on 5 Jul 2010 19:05
pk wrote in message <20100705223107.7c196afa(a)scooter.pippo.db>:
> I suppose you're wondering what kind of device a terminal is. The accurate
> answer is *very* complicated; for the purpose of your question, it can be
> seen as a device driver in the OS kernel that knows how to read and
> interpret your keystrokes and to write to the output device, be it a real
> physical terminal (eg connected to a serial port) or a virtual one (for
> example a console or an xterm).

The ":0.0" in the output seems to indicate the OP is using some kind of X11
terminal emulator.

In that case, this is what you called a "virtual one", and the exact term is
a pseudo-terminal. The principle is not very complex: a pseudo-terminal is a
kind of bidirectional pipe with some extra features. What is written on the
pseudo-terminal is read by xterm on its end of the pipe, and xterm displays
it to the window.
 |  Next  |  Last
Pages: 1 2
Prev: New proxy site 100% free
Next: does pthread on Linux use clone()?