Re: Incorrect alias expansion within command substitution

[Chet Ramey]

On 2/2/22 8:18 PM, L A Walsh wrote:

> However w/r/t starting
> with aliases being enabled by default when bash starts (interactive or not),
> I would prefer bash follow posix rules.

You have a preference that can easily be satisfied by running bash with the
appropriate option enabled. It's not a problem for interactive shells; it's
only a problem for scripts, and you have multiple options for dealing with
that.


> While I compile my bash to follow posix rules, I can't quite write my
> general scripts to expect that as bash at the trunc level

What does `trunc level' mean?


> My posix non-conformance issue has to do with bash not starting with
> aliases enabled by default in all default invocations.

Yep, that's one of the documented differences between default and posix
modes. Bash has never, and I mean never, going all the way back to 1988,
expanded aliases in non-interactive shells by default.

> While BASH_ALIASES is inherited
What does this mean?


> For that matter I can't expect my own maps (arrays with non-integer or integer
> to work in child processes.

True, bash has never exported array variables.

> I've tried to suggest various improvements over the years, and don't
> understand the resistance of all the suggestion.

Feature suggestions (not bug fixes) are evaluated against general
usability, implementation effort, and ongoing maintenance work. The time
spent implementing something -- or incorporating donated code -- has to
balance the benefit and resources I have available.

--
``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/