Eric Blake <[EMAIL PROTECTED]> writes:
> should we be using $as_expr, with a sed fallback, as in
> AS_BASENAME?
Earlier parts of our code don't, so I assume it's OK here too.
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Paul Eggert on 10/5/2007 2:53 PM:
>> + case $ac_val in
>> +/ | // ) ;;
>> +*/ ) eval $ac_var='`echo "$ac_val" | sed "s|/*\$||"`' ;;
>> + esac
>
> This mishandles ///, no?
For --prefix and --exec-prefix, no (the empty prefix is
Eric Blake <[EMAIL PROTECTED]> writes:
> + # Remove trailing slashes.
> + case $ac_val in
> +/ | // ) ;;
> +*/ ) eval $ac_var='`echo "$ac_val" | sed "s|/*\$||"`' ;;
> + esac
This mishandles ///, no? Also, 'expr' is more reliable in the
presence of weird characters, such as backslashes
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to BJörn Lindqvist on 9/29/2007 7:52 PM:
> In some autotooled project, when I type:
>
> ./configure --prefix=/usr/
>
> I want autoconf to treat it as if I typed:
>
> ./configure --prefix=/usr
>
> The fix for this problem is, as expected,