[petsc-dev] need C++ expert help

[Barry Smith]

> On Sep 13, 2016, at 10:54 PM, Jed Brown <jed at jedbrown.org> wrote:
> 
> Barry Smith <bsmith at mcs.anl.gov> writes:
> 
>>   The problem is ssize_t is not portable to Microsoft.
> 
> Is that still true?

   Yes, verified yesterday

> 
> Note that the example is buggy in the first place -- write can return
> successfully after writing less than all the bytes so you typically need
> to call it in a loop.
> 
> Why not make this example use fwrite -- it's portable and does what you want.

   Feel free.

> 
>>> On Sep 13, 2016, at 3:05 PM, Munson, Todd <tmunson at mcs.anl.gov> wrote:
>>> 
>>> 
>>> The write statement return type is ssize_t.  For whatever reason its a signed type
>>> because it can return -1 on error.
>>> 
>>> I would suggest using the ssize_t type for both and changing the "!=" to a "<"
>>> which will abort if write either returns a negative or the number of bytes 
>>> written is not correct.
>>> 
>>> Note: you will have problems if (ssize_t) (m*n*sizeof(double)) is negative 
>>> though, but that can be checked as well.
>>> 
>>> Todd.
>>> 
>>>> On Sep 13, 2016, at 2:50 PM, Satish Balay <balay at mcs.anl.gov> wrote:
>>>> 
>>>> This works..
>>>> 
>>>> Satish
>>>> 
>>>> ---------
>>>> 
>>>> diff --git a/src/mat/examples/tests/ex45.c b/src/mat/examples/tests/ex45.c
>>>> index 63d0a30..37f68e3 100644
>>>> --- a/src/mat/examples/tests/ex45.c
>>>> +++ b/src/mat/examples/tests/ex45.c
>>>> @@ -72,7 +72,7 @@ void Store2DArray(int m,int n,double *a,const char *filename,int *fdd)
>>>>     vals[i+m*j] = a[j+i*n];
>>>>   }
>>>> }
>>>> -  if (write(fd,vals,m*n*sizeof(double)) != ((unsigned int) (m*n))*sizeof(double)) abort();
>>>> +  if ((size_t)write(fd,vals,m*n*sizeof(double)) != (size_t)(m*n*sizeof(double))) abort();
>>>> free(vals);
>>>> 
>>>> }
>>>> @@ -94,7 +94,7 @@ void Store1DArray(int m,double *a,const char *filename,int *fdd)
>>>> }
>>>> if (write(fd,&classid,sizeof(int)) != sizeof(int)) abort();
>>>> if (write(fd,&m,sizeof(int)) != sizeof(int)) abort();
>>>> -  if (write(fd,a,m*sizeof(double)) != ((unsigned int)m)*sizeof(double)) abort();
>>>> +  if ((size_t)write(fd,a,m*sizeof(double)) != (size_t)(m*sizeof(double))) abort();
>>>> }
>>>> 
>>>> 
>>>> On Tue, 13 Sep 2016, Satish Balay wrote:
>>>> 
>>>>> I would tink if its typecast to same type - comparision would work..
>>>>> 
>>>>> if ((ssize_t)write(fd,vals,m*n*sizeof(double)) != (ssize_t)(m*n*sizeof(double))) abort();
>>>>> 
>>>>> I'll give this a try...
>>>>> 
>>>>> Satish
>>>>> 
>>>>> On Tue, 13 Sep 2016, Barry Smith wrote:
>>>>> 
>>>>>> 
>>>>>> I need help from someone who knows the subtleties of c++.  We have a PETSc example that always compiles fine with C but generates annoying warnings with C++. I've tried many different variants to eliminate the warnings with no luck
>>>>>> 
>>>>>> Any help is appreciated.
>>>>>> 
>>>>>>  Barry
>>>>>> 
>>>>>> 
>>>>>> mpicxx -o ex45.o -c -Wall -Wwrite-strings -Wno-strict-aliasing -Wno-unknown-pragmas -fvisibility=hidden -g -O   -fPIC    -I/sandbox/petsc/petsc.clone-3/include -I/sandbox/petsc/petsc.clone-3/arch-linux-opt-cxx-quad/include -I/usr/include/mpich2    `pwd`/ex45.c
>>>>>> /sandbox/petsc/petsc.clone-3/src/mat/examples/tests/ex45.c: In function ‘void Store2DArray(int, int, double*, const char*, int*)’:
>>>>>> /sandbox/petsc/petsc.clone-3/src/mat/examples/tests/ex45.c:75:41: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
>>>>>> if (write(fd,vals,m*n*sizeof(double)) != ((unsigned int) (m*n))*sizeof(double)) abort();
>>>>>>                                       ^
>>>>>> /sandbox/petsc/petsc.clone-3/src/mat/examples/tests/ex45.c: In function ‘void Store1DArray(int, double*, const char*, int*)’:
>>>>>> /sandbox/petsc/petsc.clone-3/src/mat/examples/tests/ex45.c:97:36: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
>>>>>> if (write(fd,a,m*sizeof(double)) != ((unsigned int)m)*sizeof(double)) abort();
>>>>>> 
>>>>>> 
>>>>>> It can be found here ftp://ftp.mcs.anl.gov/pub/petsc/nightlylogs/archive/2016/09/13/examples_next_arch-linux-opt-cxx-quad_grind.log
>>>>> 
>>>