what is the difference between child/callback and fork that is used in Perl

[Zeev Atlas]

I am new to JavaScript and node.js, but I have a lot of experience in other languages (Perl for example.) Could somebody please explain or point me to documentation about what is the difference between child/callback and fork that is used in Perl. Is there any conceptual difference, any improvement or is it basically one and the same?

Thank you
ZA

[Aria Stewart]

On Jan 12, 2015, at 10:41 AM, Zeev Atlas <zat...@gmail.com> wrote:

I am new to JavaScript and node.js, but I have a lot of experience in other languages (Perl for example.) Could somebody please explain or point me to documentation about what is the difference between child/callback and fork that is used in Perl. Is there any conceptual difference, any improvement or is it basically one and the same

Perl's fork starts a whole new unix process; communication with the parent is only via a status code.

Callbacks are a much more internal structure; they're like a sub reference in perl, passed to a sub that starts IO to be called when the IO has finished. 

Aria

[Ryan Schmidt]

On Jan 12, 2015, at 10:05 AM, Aria Stewart wrote:

But of course you *can* start a whole new unix process in node if you want to, using child_process:

http://nodejs.org/api/child_process.html

[// ravi]

On Jan 12, 2015, at 10:41 AM, Zeev Atlas <zat...@gmail.com> wrote:
>
> I am new to JavaScript and node.js, but I have a lot of experience in other languages (Perl for example.) Could somebody please explain or point me to documentation about what is the difference between child/callback and fork that is used in Perl. Is there any conceptual difference, any improvement or is it basically one and the same?
>

“child” and “callback” are different things. Can you clarify your question a bit?

—ravi

[Zeev Atlas]

The sample code I've seen is starting child process and giving it a reference to a callback function, so I perceived it as something like a 'fork' plus 'and once you finish, please call back' which is an ingenious idea but conceptually it is a fork, hence my question.

Is there any other use for callback mechanism?

Thanks

ZA 

[Aria Stewart]

> On Jan 12, 2015, at 12:29 PM, Zeev Atlas <zat...@gmail.com> wrote:
>
> The sample code I've seen is starting child process and giving it a reference to a callback function, so I perceived it as something like a 'fork' plus 'and once you finish, please call back' which is an ingenious idea but conceptually it is a fork, hence my question.
> Is there any other use for callback mechanism?

Yeah. The difference is in the separation of the task: a unix process shares nothing, and returns nothing but a status code. You can do more with plumbing some streams together, but that's about the crux of it.

With callbacks, this is more integrated in the runtime: The operations running and the callbacks all execute in the same process, and can share access to variables.

Conceptually, at 30,000 feet there are a lot of similarities, but as you get closer, the differences become very apparent.

Aria

[Ryan Schmidt]

On Jan 12, 2015, at 11:29 AM, Zeev Atlas wrote:
>
> The sample code I've seen is starting child process and giving it a reference to a callback function, so I perceived it as something like a 'fork' plus 'and once you finish, please call back' which is an ingenious idea but conceptually it is a fork, hence my question.
> Is there any other use for callback mechanism?

Callbacks are used constantly in node. You probably won't write much code in node that doesn't use a callback in some way. You tell node to do something, and give it a callback function to call when it's done doing it.

Spawning a child process is done only in special situations. For example, maybe you have a command line program you want to run to get its output. So you tell node to spawn the child process and call a callback with the output when it's done running it.

[Matt]

On Mon, Jan 12, 2015 at 12:34 PM, Aria Stewart <ared...@nbtsc.org> wrote:

The difference is in the separation of the task: a unix process shares nothing, and returns nothing but a status code

That's not true at all of fork though. It shares everything copy-on-write, and so you can communicate with the parent using a shared filehandle (a pipe), and all variable values are shared (effectively copied) at fork time. Perl makes this communication really easy by opening the magic "|-" filename (see the perlipc man page).

In node's child_process this is done slightly differently - it's not a fork, so variables aren't shared, but you do communicate over a pipe (hidden as an EventHandler). If you want to share any variables you need to send them to the child process over this pipe.