From: Omar Polo <op@omarpolo.com>
Subject: Re: regression/cmdline/diff.sh: portability fix
To: Christian Weisgerber <naddy@mips.inka.de>
Cc: gameoftrees@openbsd.org
Date: Tue, 17 Jan 2023 23:00:15 +0100

On 2023/01/17 22:32:51 +0100, Christian Weisgerber <naddy@mips.inka.de> wrote:
> The typeset shell builtin is a ksh(1) feature.  It is not available
> in a pure POSIX sh(1).  We can use printf(1) to portably perform
> the same truncation and padding of a string.
> 
> ok?

sure, ok op@.  IMHO printf(1) is even more clear than typeset here.
It's also the only usage of typeset in the regress test.

> diff /home/naddy/got
> commit - f4e8c21cb2dc6a468bae32a4dcf3a9e18f269527
> path + /home/naddy/got
> blob - 34b8116a11c50dad2f634dc6fe4dcea09570a334
> file + regress/cmdline/diff.sh
> --- regress/cmdline/diff.sh
> +++ regress/cmdline/diff.sh
> @@ -1762,10 +1762,8 @@ EOF
>  		return 1
>  	fi
>  
> -	typeset -L10 short_alpha_id
> -	typeset -L10 short_alpha_new_id
> -	short_alpha_id=$alpha_blobid
> -	short_alpha_new_id=$alpha_new_blobid
> +	short_alpha_id=$(printf '%-10.10s' $alpha_blobid)
> +	short_alpha_new_id=$(printf '%-10.10s' $alpha_new_blobid)
>  	cat <<EOF >$testroot/stdout.expected
>  diffstat $alpha_blobid $alpha_new_blobid
>   M  $short_alpha_id -> $short_alpha_new_id  |  1+  1-