Re: GNUism in groff tests

[Ingo Schwarze]

Hi Branden,

G. Branden Robinson wrote on Tue, Dec 31, 2019 at 04:25:36PM +1100:
> At 2019-12-30T18:12:58+0100, Ingo Schwarze wrote:

>> +test "X$actual" = "X$expected" && exit 0

> test "$actual" = "$expected" is entirely adequate, and POSIX-compliant.
> Adding string literals is a bodge to work around broken old pre-POSIX
> shells (or pre-POSIX test(1)).

You are right, i didn't realize "X$" is an old idiom from per-POSIX
times and not needed with POSIX shells.


>> +echo "expected: $expected"
>> +echo "actual:   $actual"
>> +exit 1

> diff output sure is easier to read.  I regret losing that.
> For the next three files I have the same two foregoing comments.
 
>> --- a/src/roff/groff/tests/string_case_xform_errors.sh
>> --- a/src/roff/groff/tests/string_case_xform_requests.sh
>> --- a/src/roff/groff/tests/string_case_xform_unicode_escape.sh

I did not commit these yet such that we can make up our mind
how to best fix them.

I admit <() is elegant, but unfortunately non-portable.


>> --- a/tmac/tests/an-old_CS_register_off.sh
>> --- a/tmac/tests/an-old_CS_register_on.sh
>> --- a/tmac/tests/an-old_CS_register_unspecified.sh
>> --- a/tmac/tests/an-old_CT_register_off.sh
>> --- a/tmac/tests/an-old_CT_register_on.sh
>> --- a/tmac/tests/an-old_CT_register_unspecified.sh

> +1.

I pushed these six to have them out of the way.
The change is so minor and trivial that i refrained from
adding a ChangeLog entry.

Yours,
  Ingo