On Wed, February 2, 2005 10:01 am, Boyd Stephen Smith Jr. said:
> On Wednesday 02 February 2005 11:59 am, "John Myers"
> <[EMAIL PROTECTED]> wrote:
>> On Wed, February 2, 2005 1:24 am, Sebastian Flothow said:
>> > Am 2. Feb 2005 um 06:35 Uhr schrieb John Myers:
>> >> It would keep the connection to the master open, and would also have
>> >> a consistent PID (unlike a shell script, which, AFAIK, may not).
>> >
>> > A shell script itself has a consistent PID. However, any command
>> > called within a shell script is assigned a new PID.
>>
>> That's not entirely true. Subshells are not executed in the same process
>> as the script.
>
> Right, but a subshell *is* another command.  It can be just as easily run
> from a prompt.
Yes, it is, but it is also part of the original line

>
>> It won't
>> work as variables are substituted by the original shell. Try this
>> instead: bash -c '$(echo $(echo $(echo $(echo $(echo $(read))))))'
>> suspending it, looking in ps, and counting how many shells you now have
>> running. (the bash -c part is so that you _can_ suspend it)
>
> Or, look at a deeply nested recursive make run by emerge in a bash inside
> su - from bash in an xterm in kde. >:)
>
> init-kdeinit-xterm-bash-su-bash-emerge-ebuild-make-make-make-make-sh-gcc
>                                                                     +as
>
I know that, but one might wish to be able to use the commands to
manipulate the progress data inside a subshell. Thus, we cannot rely on
PIDs alone. I can't think of such a situation off the top of my head, but
it would be silly to ignore it. Also, one might even need to call the
connect command from inside a subshell for some reason.


--
[email protected] mailing list

Reply via email to