Re: What type for iterator of for-loops?

Hi Daniel,

thanks for your answer.

Daniel Rentz schrieb:
> Am 10.05.2011 17:48, schrieb Regina Henschel:
>> Hi,
>>
>> I'm not well-versed in C++, so I need some help in programming in
>> Splines.cxx in chart2.
>>
>> There are already
>> typedef ::std::pair< double, double > tPointType;
>> typedef ::std::vector< tPointType > tPointVecType;
>> typedef tPointVecType::size_type lcl_tSizeType;
>>
>> My data is in
>> tPointVecType aPointsIn;
>> and the last valid index is
>> lcl_tSizeType n;
>>
>> Now I need a lot of for-loops from 0 to n. They iterate over a matrix of
>> elements of type double, over an array of elements of type double, and
>> over a vector of type tPointVecType, sometimes in the same loop. What is
>> the correct type for the iterator? Should it be lcl_tSizeType or
>> sal_uInt32? What is the implication of one or the other?
>
> The correct index type for vectors is lcl_tSizeType here, which is
> typedefed to "size_t" by the vector class. On 32-bit systems, this is an
> unsigned 32-bit type, but on 64-bit systems, it is an unsigned 64-bit
> type. Using sal_uInt32 may cause compiler warnings on these systems,
> e.g. when using
>
> sal_uInt32 i = n;

Oh. That's important. I wouldn't have noticed it on my 32bit Windows and 
VC Express. I have used some of those mixed assignments and get no 
warnings. I have to revise the code in Splines.cxx

>
>
> I recommend to use real iterators. If your "n" contains the size of the
> vector, i.e. you want to iterate all elements of the vector:
>
> tPointVecType aPointsIn;
> ...
> for (tPointVecType::const_iterator it = aPointsIn.begin(),
> it_end = aPointsIn.end(); it != it_end; ++it)
> {
> /* "it" points to a pair of double */
>
> it->first; // this is the first value of the pair
> it->second; // this is the second value of the pair
> }
>
>
>> n can be as large as the count of Spreadsheet rows, theoretically, but
>> will be smaller than hundred in most cases.
>
> If "n" is less than aPointsIn.size(), you can modify the loop header:
>
> for (tPointVecType::const_iterator it = aPointsIn.begin(),
> it_end = it + n; it != it_end; ++it)
>

OK, that's clear. But I need to iterate over vector and matrix parallel 
and therefore need an unsigned integer as index for the matrix.

>
> If you still prefer iterating by index: :-)
>
> for (lcl_tSizeType i = 0; i < n; ++i)
> {
> aPointsIn[i].first;
> aPointsIn[i].second;
> }

I think, that works best for me. I have to solve a LGS with a vector on 
right side and a matrix, which is nearly, but not exact, banded. So I 
need some index calculations for shifting and for calculating, whether I 
can terminate the loops earlier.

Kind regards
Regina


-- 
-----------------------------------------------------------------
To unsubscribe send email to [email protected]
For additional commands send email to [email protected]
with Subject: help