Requiring CPP11 by default in DGtal

[dcoeurjo]

This PR:

• remove some #ifdef WITH_C11 switches and update the code accordingly
• change some deprecated boost::array to std::array
• various cpp11 related fixes

[dcoeurjo]

I'm still working on it, it just wanted to notify the devs.

[rolanddenis]

👍

[dcoeurjo]

@rolanddenis I have an issue with catch on travis (see https://gist.github.com/dcoeurjo/888095e6f971e2d25267)... any idea?

[rolanddenis]

It seems that gcc doesn't support std::underlying_type until version 4.7 (see eg http://stackoverflow.com/questions/7931378/how-do-i-get-the-fundamental-type-of-an-enum)
A fix could be to define CATCH_CONFIG_CPP11_NO_IS_ENUM when a previous gcc version is used.

[dcoeurjo]

@JacquesOlivierLachaud : std::binder1st is deprecated in c++11 (see LocalEstimatorFromSurfelFunctorAdapter.h, line 145) and an error is raised on travis.

Coud you please have a look to the code ? It is unclear for me how to fix it with C++11 stuffs (better solution than redefining a "binder" inner struct)

[rolanddenis]

A way to solve this is to replace
typedef std::binder1st<Metric> MetricToPoint;
by
typedef std::function< typename Metric::Value ( typename Metric::Point ) > MetricToPoint;
including <functional> header.

In LocalEstimatorFromSurfelFunctorAdapter.ih, line 170, the construction of metricToPoint becomes:
MetricToPoint metricToPoint = std::bind( *myMetric, myEmbedder( *it ), std::placeholders::_1 );
(adding constness to metricToPoint and vfunctor is possible).

It is also possible to directly construct vfunctor from a lambda and thus avoiding declaring those private typedef.

Note: the documentation talks about a CMetric concept, is it not rather CMetricSpace ?

[dcoeurjo]

Thanks @rolanddenis !

[dcoeurjo]

(Yeah.. CMetric -> CMetricSpace, fixed now)

[dcoeurjo]

Allright, this PR is ready for review.

[kerautret]

I look it and also with DGtalTools

[rolanddenis]

The "WITH_C11" value is still displayed and "false".
See https:/dcoeurjo/DGtal/blob/cpp11/cmake/CheckDGtalOptionalDependencies.cmake#L39

[dcoeurjo]

you're right. I'll clean this up

[kerautret]

since we can no more use it without c++11 we can update "or disable this WITH_C11 option"

[kerautret]

looks fine for me, tested on different compilators, and it looks also to fix the DGtalTools "rand" related bug. ;) !
Just correct also the message that appears on non cpp11 compilers.
And also the main documentation we should precise it: (in moduleBuildDGtal.dox) ?
(and also here http://dgtal.org/download/)

[dcoeurjo]

yep, you're right.

[dcoeurjo]

I've also changed the dgtal.org web page.
Ready for a merge ?

[kerautret]

yes thanks, perfect ! I have also the fix in DGtalTools, merging..

[dcoeurjo]

Thanks... Please have a look to #1090 for the boost::random cleanup.