Re: [Bug-apl] SVN 839 doesn't compile

[Xiao-Yong Jin]

Looks like a gcc loop unrolling bug.

I have another issue.

cc1plus: warnings being treated as errors                                       
                                                                                
                                            In file included from 
Quad_SQL.cc:27:                                                                 
                                                                                
                      sql/apl-sqlite.hh:24: error: expected 
[error|warning|ignored] after '#pragma GCC diagnostic'

The diagnostic push/pop pragma was introduced in gcc 4.6.

> On Jan 16, 2017, at 3:12 PM, Juergen Sauermann <address@hidden> wrote:
> 
> Hi Xiao-Yong,
> 
> I believe the warning is equally bogus as the one below.
> 
> It is true that other.rho_rho is int32_t while Shapeitem is int64_t, but
> the comparison in the loop() macto is between two int64_ts. And has
> nothing to do with array subscripts.
> 
> /// Jürgen
> 
> 
> On 01/16/2017 07:18 PM, Xiao-Yong Jin wrote:
>> I'm seeing the same array subscript errors with gcc 4.4.7 here.
>> 
>> g++ -DHAVE_CONFIG_H -I. -I..    -Werror -Wall -I sql -I/usr/include   
>> -rdynamic -mcpu=power7 -mtune=power7 -O3 -fno-math-errno 
>> -funsafe-math-optimizations -fno-rounding-math -fno-signaling-nans 
>> -fcx-limited-range -pipe -MT apl-ArrayIterator.o -MD -MP -MF 
>> .deps/apl-ArrayIterator.Tpo -c -o apl-ArrayIterator.o `test -f 
>> 'ArrayIterator.cc' || echo './'`ArrayIterator.cc
>> cc1plus: warnings being treated as errors
>> Shape.hh: In constructor 'ArrayIteratorBase::ArrayIteratorBase(const 
>> Shape&)':
>> Shape.hh:66: error: array subscript is above array bounds
>> make[3]: *** [apl-ArrayIterator.o] Error 1
>> 
>> The line 66 of Shape.hh reads
>> 62    /// constructor: shape of another shape
>> 63    Shape(const Shape & other)
>> 64    : rho_rho(0),
>> 65      volume(1)
>> 66    { loop(r, other.rho_rho)   add_shape_item(other.rho[r]); }
>> 
>> The loop is a macro in Common.hh
>> #define loop(v, e) for (ShapeItem v = 0; v < ShapeItem(e); ++v)
>> 
>> The warning (treated as array) is because ShapeItem (v) is int64_t, while 
>> Rank (other.rho_rho) is int32_t.
>> src/APL_types.hh:42:typedef int32_t Rank;
>> src/APL_types.hh:48:typedef int64_t ShapeItem;
>> 
>> 
>> 
>>> On Jan 11, 2017, at 10:08 AM, Juergen Sauermann <address@hidden>
>>>  wrote:
>>> 
>>> Hi,
>>> 
>>> the warnings should be fixed now, SVN 844.
>>> 
>>> The warning on raspberry pi:
>>> 
>>> Shape.hh: In member function ‘virtual Token 
>>> Bif_F12_FORMAT::eval_B(Value_P)’: 
>>> Shape.hh:125:18: error: array subscript is below array bounds 
>>> [-Werror=array-bounds] 
>>>          if (rho[r])   { volume /= rho[r];  rho[r] = sh;  volume *= rho[r]; 
>>> } 
>>>                   ^ 
>>> Shape.hh:126:30: error: array subscript is below array bounds 
>>> [-Werror=array-bounds] 
>>>          else          { rho[r] = sh;   recompute_volume();                 
>>> } } 
>>>                               ^ 
>>> 
>>> seems to be entirely bogus. I have checked all occurrences of 
>>> Shape::set_shape_item(), including
>>> all inlined functions in Shape.hh using Shape::set_shape_item(), in the 
>>> functions mentioned, and I
>>> could not find any reason while the compiler came to that conclusion. 
>>> Unfortunately the compiler did
>>> not bother to tell at which line number this happens.
>>> 
>>> I also updated FILE_IO.apl as proposed.
>>> 
>>> /// Jürgen
>>> 
>>> 
>>> 
>>> On 01/11/2017 06:47 AM, David B. Lamkins wrote:
>>> 
>>>> There's one remaining when compiling 843 under clang. The attached patch 
>>>> fixes that as well as the remaining error reported by gcc.
>>>> 
>>>> 
>>>> 
>>>> On Tue, Jan 10, 2017 at 03:54:36PM -0800, David Lamkins wrote:
>>>> 
>>>> 
>>>>> Worth noting: There are now no warnings or notes when compiling under gcc 
>>>>> 6.3.1.
>>>>> 
>>>>> On Tue, Jan 10, 2017 at 3:45 PM, David Lamkins 
>>>>> 
>>>>> <address@hidden>
>>>>> 
>>>>>  wrote:
>>>>> 
>>>>> 
>>>>>> Thank you, Juergen.
>>>>>> 
>>>>>> SVN 843 compiles under gcc 6.3.1.
>>>>>> 
>>>>>> This is the sole remaining error when compiling with clang 3.8.0:
>>>>>> 
>>>>>> clang++ -DHAVE_CONFIG_H -I. -I..    -Werror -Wall -Wno-strict-aliasing
>>>>>> -Wno-deprecated-declarations -I/usr/include -I sql  -I/usr/include -I
>>>>>> sql   -g -O2 -MT apl-UserFunction_header.o -MD -MP -MF
>>>>>> .deps/apl-UserFunction_header.Tpo -c
>>>>>>  -o apl-UserFunction_header.o `test -f 'UserFunction_header.cc' ||
>>>>>> echo './'`UserFunction_header.cc
>>>>>> Tokenizer.cc:604:42: error: explicitly assigning value of variable of
>>>>>> type 'APL_Float' (aka 'double') to itself [-Werror,-Wself-assign]
>>>>>>         if (!imag_need_float)   imag_flt = imag_flt;
>>>>>>                                 ~~~~~~~~ ^ ~~~~~~~~
>>>>>> Tokenizer.cc:637:45: error: explicitly assigning value of variable of
>>>>>> type 'APL_Float' (aka 'double') to itself [-Werror,-Wself-assign]
>>>>>>         if (!imag_need_float)   degrees_flt = degrees_flt;
>>>>>>                                 ~~~~~~~~~~~ ^ ~~~~~~~~~~~
>>>>>> 2 errors generated.
>>>>>> 
>>>>>> 
>>>>>> On Tue, Jan 10, 2017 at 11:54 AM, Juergen Sauermann
>>>>>> 
>>>>>> 
>>>>>> <address@hidden>
>>>>>> 
>>>>>>  wrote:
>>>>>> 
>>>>>> 
>>>>>>> Hi,
>>>>>>> 
>>>>>>> I believe that I fixed most warnings, *SVN 842*.
>>>>>>> 
>>>>>>> Since there were many warnings and from different sources, I might have
>>>>>>> overlooked
>>>>>>> some of them (or introduced new ones). Please keep reporting them.
>>>>>>> 
>>>>>>> /// Jürgen
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>> 
>>>>>> --
>>>>>> "I think as musicians, we have more vision than we credit ourselves
>>>>>> for. We should dive into it without reserve, and take the sort of
>>>>>> risks it takes to be divisive and impactful. Things have yet to be
>>>>>> done."
>>>>>>    Aaron Fowler Clark
>>>>>> 
>>>>>> "Do not seek to follow in the footsteps of the masters. Seek what they 
>>>>>> sought."
>>>>>>    Matsuo Basho
>>>>>> 
>>>>>> "Knowledge speaks; wisdom listens."
>>>>>>    Jimi Hendrix
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> http://soundcloud.com/davidlamkins
>>>>>> http://reverbnation.com/lamkins
>>>>>> http://reverbnation.com/lcw
>>>>>> http://lamkins-guitar.com/
>>>>>> http://lamkins.net/
>>>>>> http://successful-lisp.com/
>>>>> 
>>>>> -- 
>>>>> "I think as musicians, we have more vision than we credit ourselves
>>>>> for. We should dive into it without reserve, and take the sort of
>>>>> risks it takes to be divisive and impactful. Things have yet to be
>>>>> done."
>>>>>    Aaron Fowler Clark
>>>>> 
>>>>> "Do not seek to follow in the footsteps of the masters. Seek what they 
>>>>> sought."
>>>>>    Matsuo Basho
>>>>> 
>>>>> "Knowledge speaks; wisdom listens."
>>>>>    Jimi Hendrix
>>>>> 
>>>>> 
>>>>> 
>>>>> http://soundcloud.com/davidlamkins
>>>>> http://reverbnation.com/lamkins
>>>>> http://reverbnation.com/lcw
>>>>> http://lamkins-guitar.com/
>>>>> http://lamkins.net/
>>>>> http://successful-lisp.com/
>> 
>