Local variables for nested function calls

[eduardo-chibas]

> Sent: Thursday, July 29, 2021 at 6:59 AM
> From: "Greg Wooledge" <greg@wooledge.org>
> To: help-bash@gnu.org
> Subject: Re: Local variables for nested function calls
>
> On Wed, Jul 28, 2021 at 08:46:56PM +0200, eduardo-chibas@caramail.com wrote:
> > The problem started with a function `plist` that calls another function 
> > `region`.
> >
> > ----- function plist -----
> >
> > ropts=("$@")
> >
> > # read function arguments here
> > "-d")
> >   local -r dpath="$2" ; shift 2 ;;
> >
> > local fdir=${dpath:-$PWD}
> >
> > region "${ropts[@]}"
> >
> > ----- plist ends here -----
>
> You have ropts as non-local variable here, FYI.  That may or may not
> be what you want.
>
> > ----- function region -----
> >
> > # read function arguments here
> >  "-d")
> >     local fdir=$2 ; shift 2 ;;
> >
> > : "${fdir:="${@:$#}"}"}
> >
> > ----- region ends here -----
>
> > When I was calling `plist` without the -d option, and subsequently calling
> > `region`, the value of `fdir` was being set to `$PWD`, rather than the last
> > argument passed to `region`.
>
> If you don't include the -d option, then you're skipping over the
> "local fdir" in the second function, which means it's going to use dynamic
> scope to find the fdir variable in the first function.
>
> > This means that I have to declare my variables local without setting them
> > first.
>
> Well, yeah.  If region is supposed to have its own local fdir which is
> independent of the one in plist, then you should declare fdir in region.
> And not only when the -d option is given.
>
> As a general programming principle, you've got a block of code which
> has a local variable defined inside it, which is great.  But then
> you're using the variable after that block of code has terminated.
> That's *not* great.  In some languages, that would be an error (the
> compiler or interpreter would yell at you).  In bash, it's not an error;
> you have to catch it yourself.

When I use

local this that

`this` and `that` are unset as expected.  At first I thought the command

that=${this:-8}

would not set the default because `this` was defined as local.


> If fdir is supposed to be local throughout the entire function, then I'd
> declare it at the *top* of the function.  Or as close to the top as you
> can reasonably make it.
>
> > But then I see no point of using parameter expansion such as `: 
> > "${fdir:="${@:$#}"}"}`
> > to set defaults because `fdir` would already be defined when I use the 
> > local declaration
> > within `region`.
>
> ... I still don't know what you're trying to *do*, though.
>
> I don't know what "fdir" even stands for, or why you're setting it, or
> why you're setting it a second time, or how you use it.
>
> That makes it hard to give advice beyond generalities.
>
>