Re: Incorrect alias expansion within command substitution

[Chet Ramey]

On 2/3/22 5:25 PM, Robert Elz wrote:
>      Date:        Thu, 3 Feb 2022 16:19:08 -0500
>      From:        Greg Wooledge <greg@wooledge.org>
>      Message-ID:  <YfxGzCsd4AgoKTM/@wooledge.org>
>
>    | This declares the variables at the wrong scope.
>
> You are correct - I never normally use "declare" for anything, so
> I had forgotten that it was essentially a clone of "local" (rather
> than "local" being "declare -l" or something similar (I know
> declare has a "-l" which is something entirely different, that's
> not the point).

That has the cart before the horse. `local' is a synonym for `declare'.
`declare' has -- and has always had -- the ksh `typeset' semantics: if you
use it in a shell function, you get a local variable. `local' exists
because Posix 1003.2 draft 9 included it as a builtin, and we already had
the same functionality in `declare', so `local' came in as a synonym with
a check to make sure you were currently executing a shell function. Of
course, on the next draft, local variables were thrown out entirely.

> Since this is bash local stuff, it would be useful to add an "uplevel"
> option to declare, so that things like this can work.   It has -g already,
> and while useful, it doesn't solve this problem.

Interesting idea. It would be a no-op (or an error) at the global scope,
naturally.


--
``The lyf so short, the craft so long to lerne.'' - Chaucer
                 ``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, UTech, CWRU    chet@case.edu    http://tiswww.cwru.edu/~chet/