[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: GDAL 2.3.0



On samedi 21 avril 2018 19:58:42 CEST Sebastiaan Couwenberg wrote:

> On 04/21/2018 07:53 PM, Even Rouault wrote:

> > On samedi 21 avril 2018 19:32:23 CEST Sebastiaan Couwenberg wrote:

> >> Hi Even,

> >>

> >> Regarding the C++11 requirement, this causes the liblas build to fail

> >> due to the check in cpl_port.h.

> >>

> >> liblas uses -std=c++98 whereas the gdal build used -std=gnu89

> >

> > You mean GDAL uses -std=c++11 right ?

>

> No, the build used -std=gnu89, see the attached build log.

 

Ah, std=gnu89 is used only to compiled .c files (to make sure we don't use a higher C dialect, that wouldn't be supported by MSVC)

But C++ files are compiled without any -std= since presumably the compiler you use on debian unstable defaults to C++14

 

>

> >> . Do you

> >> recommend the latter for projects that depend on GDAL? Or should they

> >> prefer -std=c++11?

> >

> > Projects using GDAL C++ API should use at least -std=c++11 (or higher)

>

> See my previous message in this thread. Just changing -std=c++98 to

> -std=c++11 did not fix the liblas build failure.

 

Hum weird. I've just managed to build libLAS 1.8.1 against GDAL 2.3.0, with cmake, by forcing CMAKE_CXX_FLAGS_RELEASE= -O3 -DNDEBUG -std=c++11, with the gcc 5.4 of my ubuntu 16.04

 

Actually I see in the log that -ansi is used, and according to gcc manpage, -ansi in C++ mode is equivalent to -std=c++98. So perhaps -ansi -std=c++11 is equivalent to -ansi.

 

Even

 

 

--

Spatialys - Geospatial professional services

http://www.spatialys.com


Reply to: