Std. IO streams in Ruby

In my rails controller, I am doing the following :

@output = `g++ j.cpp -o "prog" && ./prog`

This gives the output in the @output variable which i can display in my
view. But the above works only if the j.cpp is correct and doesn't
expect any user input. How can I use the stdin/stderr and stdout streams
here so that :

If the user has to give input, I open a dialog box on the view with a
textfield where he/she can enter the input, and the program continues to
execute.
If there are any errors in the file, then I should be able to get the
errors and display them to the user.
I tried doing this :

@output = `g++ j.cpp -o "prog" && ./prog| tee prog`

This allows me to enter the user input at the server command prompt (the
server log...i don't know what do we call it), but I want it to be
entered in a textfield in the view.

In short, how can I get control of stdin, so that whenever the stdin
waits for some user input, i get to know that it is waiting for the user
input and then i can show the user a text field and get the user input
to be passed in the stdin stream.
Let me know if there's a doubt in the question.

In my rails controller, I am doing the following :

@output = `g++ j.cpp -o "prog" && ./prog`

This gives the output in the @output variable which i can display in my
view. But the above works only if the j.cpp is correct and doesn't
expect any user input. How can I use the stdin/stderr and stdout streams
here so that :

If the user has to give input, I open a dialog box on the view with a
textfield where he/she can enter the input, and the program continues to
execute.
If there are any errors in the file, then I should be able to get the
errors and display them to the user.
I tried doing this :

@output = `g++ j.cpp -o "prog" && ./prog| tee prog`

This allows me to enter the user input at the server command prompt (the
server log...i don't know what do we call it), but I want it to be
entered in a textfield in the view.

In short, how can I get control of stdin, so that whenever the stdin
waits for some user input, i get to know that it is waiting for the user
input and then i can show the user a text field and get the user input
to be passed in the stdin stream.
Let me know if there's a doubt in the question.

Well in a plain old ruby script IO.popen allows you to run a command
and gets you the command stdin/stdout (and open3/popen3 gets you
stderr too).

Controlling this interactively via a webapp is much more tricky. A
single http request only has a single response, so you'd have to break
this up across multiple requests. There's no guarantee that the next
request from the user will go to the same rails instance, so figuring
out what do do with the IO objects connecting you to the command being
run is tricky enough.

One possible way out is to have some long running daemon process that
does that actual running of the command and have your rails actions
communicate with that. If just polling for whether the command is
waiting for input isn't acceptable then you might need to look at
comet and that sort of thing. If you can restrict yourself to specific
browsers, the new html5 websockets stuff might be interesting too

Fred

Frederick Cheung wrote:

textfield where he/she can enter the input, and the program continues to

In short, how can I get control of stdin, so that whenever the stdin
waits for some user input, i get to know that it is waiting for the user
input and then i can show the user a text field and get the user input
to be passed in the stdin stream.
Let me know if there's a doubt in the question.

Well in a plain old ruby script IO.popen allows you to run a command
and gets you the command stdin/stdout (and open3/popen3 gets you
stderr too).

Controlling this interactively via a webapp is much more tricky. A
single http request only has a single response, so you'd have to break
this up across multiple requests. There's no guarantee that the next
request from the user will go to the same rails instance, so figuring
out what do do with the IO objects connecting you to the command being
run is tricky enough.

One possible way out is to have some long running daemon process that
does that actual running of the command and have your rails actions
communicate with that. If just polling for whether the command is
waiting for input isn't acceptable then you might need to look at
comet and that sort of thing. If you can restrict yourself to specific
browsers, the new html5 websockets stuff might be interesting too

Fred

@Fred :-I have been looking over the internet for a solution but never
got a real answer. Thanks for your help.
My web app allows the user to run code in the browser window itself, so
do i really need to do polling for that? I mean is there any better
solution? If not, how can I use your approach. I have never tried
something like that before, so a little confused. Can you explain it
using a little code snippet? That would really help me out.
Also, the responses would be through AJAX. The user code will be sent
back to server through an AJAX call, the file created for the code and
then the response back to the browser(the output of file executed) in an
AJAX call. So, that guarantee that next request will go to same rails
instance, won't that vanish away or not? Could be elaborate a little bit
on this too.

Frederick Cheung wrote:
> On Aug 1, 7:29 pm, Jatin Kumar <li.

>

@Fred :-I have been looking over the internet for a solution but never
got a real answer. Thanks for your help.
My web app allows the user to run code in the browser window itself, so
do i really need to do polling for that? I mean is there any better
solution? If not, how can I use your approach. I have never tried
something like that before, so a little confused. Can you explain it
using a little code snippet? That would really help me out.

Don't have any code snippets, just one way I'd try tackling this - run
the commands from a separate ruby process and use whatever mechanism
you want (drb might be handy) to allow your controller action to get
from this process what the state of the command being run is.

Also, the responses would be through AJAX. The user code will be sent
back to server through an AJAX call, the file created for the code and
then the response back to the browser(the output of file executed) in an
AJAX call. So, that guarantee that next request will go to same rails
instance, won't that vanish away or not? Could be elaborate a little bit
on this too.

Unless I've misunderstood your command needs to keep running over the
course of several requests if you're going to be getting input from
the users at the point. Unless you're willing for your rails code to
just block until input is required ( not a great idea) you'll need
some form of polling/comet.