Re: New FAILs on gdb.fortran/short-circuit-argument-list.exp when testing on native{,-extended}-gdbserver

[Richard Bunt <Richard.Bunt@arm.com>]

On 12/3/18 1:22 PM, Richard Bunt wrote:

The assertions in the test check against the program's standard out which
appears in a different place when the test is run using gdb server. So the
functionality is fine, but the test case is flawed.

I will make a patch to fix this.

Many thanks,

Rich.

> Hi Sergio,
> 
> My apologies, I completely missed this reply. I'll investigate these failures.
> 
> Rich.
> 
> On 10/7/18 6:14 AM, Sergio Durigan Junior wrote:
>> On Tuesday, September 18 2018, Richard Bunt wrote:
>>
>>> On 09/14/2018 10:00 PM, Tom Tromey wrote:
>>>>>>>>> "Richard" == Richard Bunt <richard.bunt@arm.com> writes:
>>>>
>>>> Richard> Version 3 addresses review comments:
>>>>
>>>> Richard> * Removed forward declaration of skip_undetermined_arglist.
>>>> Richard> * Made definition of skip_undetermined_arglist static.
>>>> Richard> * Added introductory comment to skip_undetermined_arglist.
>>>> Richard> * Dropped (*pos)+=3 into the block.
>>>> Richard> * Adjusted Tcl bracing.
>>>> Richard> * Removed string splitting in the test case.
>>>>
>>>> Thanks again for the patch.
>>>>
>>>> Richard> +	if (noside != EVAL_SKIP)
>>>> Richard> +	  {
>>>> Richard> +	    return evaluate_var_value (noside, exp->elts[pc + 1].block, var);
>>>> Richard> +	  }
>>>>
>>>> The gdb style is no braces if there's just a single line.
>>>>
>>>> Richard> +	else
>>>> Richard> +	  {
>>>> Richard> +	    /* Return a dummy value of the correct type when skipping, so
>>>> Richard> +	       that parent functions know what is to be skipped.  */
>>>> Richard> +	    return allocate_value (SYMBOL_TYPE (var));
>>>>
>>>> ... compared to here, where braces are mandatory.
>>>>
>>>> This patch is ok with that first set of braces removed.
>>>> (You don't have to re-send it.)
>>>>
>>>> thanks,
>>>> Tom
>>>>
>>>
>>> Many thanks for reviewing this change. I will fix up the braces and push.
>>
>> Hi Richard,
>>
>> Just a heads up, I've noticed a few failures when running
>> gdb.fortran/short-circuit-argument-list.exp on a
>> native{,-extended}-gdbserver board:
>>
>>   FAIL: gdb.fortran/short-circuit-argument-list.exp: arg=No: p function_no_arg_false() .OR. function_no_arg()
>>   FAIL: gdb.fortran/short-circuit-argument-list.exp: arg=One: p function_no_arg_false() .OR. function_one_arg( .TRUE.)
>>   FAIL: gdb.fortran/short-circuit-argument-list.exp: arg=Two: p function_no_arg_false() .OR. function_two_arg( .TRUE., .TRUE.)
>>   FAIL: gdb.fortran/short-circuit-argument-list.exp: p function_one_arg(.FALSE. .OR. function_no_arg())
>>   FAIL: gdb.fortran/short-circuit-argument-list.exp: p function_one_arg(.TRUE. .OR. function_no_arg())
>>
>> The failures are easily reproducible, but let me know if you need a
>> gdb.log excerpt.  I haven't had the time to investigate them further.
>>
>> Thanks,
>>