[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 'local -x VARIABLE' does not clear variable for subprocesses

From: Linda Walsh
Subject: Re: 'local -x VARIABLE' does not clear variable for subprocesses
Date: Mon, 05 May 2014 15:21:30 -0700
User-agent: Thunderbird

Chet Ramey wrote:

Yes.  Once you assign a value to the local instances of the variables,
they are set.

Local variables inherit the exported status of the variable they are
shadowing to avoid surprises in the export environment.  Believe it or
not, it's less surprising than letting a local variable shadow a global
for everything except exporting to child processes.  It's been that way
forever (well, at least as far back as bash-2.05b, when I quit looking.)
I never knew that -- always made sure to re-export vars
I wanted exported...so had never had an occasion to run into
that 'feature'.

Um... is it really needed/helpful to people that way?, as
it surprises me more, to realize that a 'local' variable,
again, isn't really just 'local'.

In that regard, it's like perl, with local localizing the
value of a global while it is running (or for sub-children).

That leaves a opportunity -- "my"... lexically scoped local vars
that disappear outside of the lexical scope (unless explicitly
exported.  While that doesn't sound exceptionally useful by
itself -- it would ***also*** allow use of a global "read-only"
name in a subroutine without worrying about whether or not
your "local varname" is prohibited due to some upper level call
having made it 'read only'.

Any subs called by the 'sub' would see the namespace as it appeared
before entering the sub with the 'my' and would see the read-only
values, unchanged, as they'd be before calling the function with 'my'.

I.e. 'my' vars would only let a function use any name without
regard to what environment it was called from, but that name is
only viewable and active from within that 1 function.

I would restrict lexically scoped 'my' vars, from being exported --
as that would break the paradigm.  If a user wants to store a value
in a global or exported var with the same name as a lexical
'my-declared' var, then the lexical would first have to be "unset".
Then any previously declared globals or exported vars by that
name would be visible.

ex: this would no longer fail (infinite loop):


function printnames {
  local -i i  #should be 'lexical override'...
  echo -n "Names = "
  for ((i=0;i<address@hidden;++i));do echo -n "${names[i]} "; done
  echo ""

declare -a names=('sam' 'han' 'riker')

declare -r  i=2

read -p "go on to see loop or exit ([e],g) " var
if [[ $var != g ]]; then exit; fi


reply via email to

[Prev in Thread] Current Thread [Next in Thread]