On Sun, 21. Feb 2016 at 15:47:00 +0100, Markus Neteler wrote:
since I’m not familiar with compilation on Windows, could you briefly
indicate what’s needed / which efforts to compile GRASS GIS with MSVC rather
than mingw given the recent efforts on the 64bit packaging? Just to better
understand…
Not sure. I’d probably start with moving the build system to CMake. Like I
already did for just the GRASS libraries at some point in the past (some code
changes I submitted to make the libraries build with MSVC were also accepted
back then).
But moving all the modules there is probably a much bigger task and I’m also
not sure about the acceptance of this.
It might also be possible to integrate msvc into configure/make (at least I
think I’ve seen configures with MSVC stuff in them). Although you’d still need
a unix like shell, make and other tools from mingw/cygwin. But I never
explored that option (mainly just because the CMake approach went much better
with what I needed for QGIS).
But again GRASS works fine in OSGeo4W AFAICT (and as you know and that’s still
not much ;)) and hence I don’t see the big problem of mixing MinGW and MSVC
DLLs. It doesn’t add much to the mess of having different MSVC compilers and
runtimes around. IMHO that part of Rashad’s issue is just a CMake bug and not
a fundamental flaw in OSGeo4W.
Jürgen
–
Jürgen E. Fischer norBIT GmbH Tel. +49-4931-918175-31
Dipl.-Inf. (FH) Rheinstraße 13 Fax. +49-4931-918175-50
Software Engineer D-26506 Norden http://www.norbit.de
QGIS release manager (PSC) Germany IRC: jef on FreeNode
On Sun, 21. Feb 2016 at 15:47:00 +0100, Markus Neteler wrote:
since I’m not familiar with compilation on Windows, could you briefly
indicate what’s needed / which efforts to compile GRASS GIS with MSVC rather
than mingw given the recent efforts on the 64bit packaging? Just to better
understand…
Not sure. I’d probably start with moving the build system to CMake. Like I
already did for just the GRASS libraries at some point in the past (some code
changes I submitted to make the libraries build with MSVC were also accepted
back then).
But moving all the modules there is probably a much bigger task and I’m also
not sure about the acceptance of this.
It might also be possible to integrate msvc into configure/make (at least I
think I’ve seen configures with MSVC stuff in them). Although you’d still need
a unix like shell, make and other tools from mingw/cygwin. But I never
explored that option (mainly just because the CMake approach went much better
with what I needed for QGIS).
But again GRASS works fine in OSGeo4W AFAICT (and as you know and that’s still
not much ;)) and hence I don’t see the big problem of mixing MinGW and MSVC
DLLs. It doesn’t add much to the mess of having different MSVC compilers and
runtimes around. IMHO that part of Rashad’s issue is just a CMake bug and not
a fundamental flaw in OSGeo4W.
Jürgen
–
Jürgen E. Fischer norBIT GmbH Tel. +49-4931-918175-31
Dipl.-Inf. (FH) Rheinstraße 13 Fax. +49-4931-918175-50
Software Engineer D-26506 Norden http://www.norbit.de
QGIS release manager (PSC) Germany IRC: jef on FreeNode
On Sun, 21. Feb 2016 at 15:47:00 +0100, Markus Neteler wrote:
since I'm not familiar with compilation on Windows, could you briefly
indicate what's needed / which efforts to compile GRASS GIS with MSVC
rather
than mingw given the recent efforts on the 64bit packaging? Just to
better
understand...
Not sure. I'd probably start with moving the build system to CMake. Like I
already did for just the GRASS libraries at some point in the past (some
code
changes I submitted to make the libraries build with MSVC were also accepted
back then).
But moving all the modules there is probably a much bigger task and I'm also
not sure about the acceptance of this.
It might also be possible to integrate msvc into configure/make (at least I
think I've seen configures with MSVC stuff in them). Although you'd still
need
a unix like shell, make and other tools from mingw/cygwin. But I never
explored that option (mainly just because the CMake approach went much
better
with what I needed for QGIS).
But again GRASS works fine in OSGeo4W AFAICT (and as you know and that's
still
not much ;)) and hence I don't see the big problem of mixing MinGW and MSVC
DLLs.
Also, it is not possible to debug GRASS lib built with MinGW in an
application built with MSVC.
Radim
It doesn't add much to the mess of having different MSVC compilers
and
runtimes around. IMHO that part of Rashad's issue is just a CMake bug and
not
a fundamental flaw in OSGeo4W.
Jürgen
--
Jürgen E. Fischer norBIT GmbH Tel. +49-4931-918175-31
Dipl.-Inf. (FH) Rheinstraße 13 Fax. +49-4931-918175-50
Software Engineer D-26506 Norden http://www.norbit.de
QGIS release manager (PSC) Germany IRC: jef on FreeNode
On Sun, Feb 21, 2016 at 5:15 PM, Rashad Kanavath
<mohammedrashadkm@gmail.com> wrote:
Hello Markus,
As this part related to GRASS, I am writing it here.
Moving to cmake is one thing. But I don't know if that is blocking msvc
build. Did anybody tried to build grass with msvc?
Yes I did, the libs only.
I started with https://github.com/swig/cccl but that was failing
because cl.exe was givin deprecation warning.....
In the end, I managed to compile libs with command (manually changed Makefile):
On Sun, 21. Feb 2016 at 15:47:00 +0100, Markus Neteler wrote:
> since I'm not familiar with compilation on Windows, could you briefly
> indicate what's needed / which efforts to compile GRASS GIS with MSVC
> rather
> than mingw given the recent efforts on the 64bit packaging? Just to
> better
> understand...
Not sure. I'd probably start with moving the build system to CMake. Like
I
already did for just the GRASS libraries at some point in the past (some
code
changes I submitted to make the libraries build with MSVC were also
accepted
back then).
But moving all the modules there is probably a much bigger task and I'm
also
not sure about the acceptance of this.
It might also be possible to integrate msvc into configure/make (at least
I
think I've seen configures with MSVC stuff in them). Although you'd still
need
a unix like shell, make and other tools from mingw/cygwin. But I never
explored that option (mainly just because the CMake approach went much
better
with what I needed for QGIS).
But again GRASS works fine in OSGeo4W AFAICT (and as you know and that's
still
not much ;)) and hence I don't see the big problem of mixing MinGW and
MSVC
DLLs. It doesn't add much to the mess of having different MSVC compilers
and
runtimes around. IMHO that part of Rashad's issue is just a CMake bug
and not
a fundamental flaw in OSGeo4W.
Jürgen
--
Jürgen E. Fischer norBIT GmbH Tel.
+49-4931-918175-31
Dipl.-Inf. (FH) Rheinstraße 13 Fax.
+49-4931-918175-50
Software Engineer D-26506 Norden http://www.norbit.de
QGIS release manager (PSC) Germany IRC: jef on
FreeNode
As this part related to GRASS, I am writing it here.
Moving to cmake is one thing. But I don’t know if that is blocking msvc
build. Did anybody tried to build grass with msvc?
Yes I did, the libs only.
I started with https://github.com/swig/cccl but that was failing
because cl.exe was givin deprecation warning…
In the end, I managed to compile libs with command (manually changed Makefile):
On Sun, 21. Feb 2016 at 15:47:00 +0100, Markus Neteler wrote:
since I’m not familiar with compilation on Windows, could you briefly
indicate what’s needed / which efforts to compile GRASS GIS with MSVC
rather
than mingw given the recent efforts on the 64bit packaging? Just to
better
understand…
Not sure. I’d probably start with moving the build system to CMake. Like
I
already did for just the GRASS libraries at some point in the past (some
code
changes I submitted to make the libraries build with MSVC were also
accepted
back then).
But moving all the modules there is probably a much bigger task and I’m
also
not sure about the acceptance of this.
It might also be possible to integrate msvc into configure/make (at least
I
think I’ve seen configures with MSVC stuff in them). Although you’d still
need
a unix like shell, make and other tools from mingw/cygwin. But I never
explored that option (mainly just because the CMake approach went much
better
with what I needed for QGIS).
But again GRASS works fine in OSGeo4W AFAICT (and as you know and that’s
still
not much ;)) and hence I don’t see the big problem of mixing MinGW and
MSVC
DLLs. It doesn’t add much to the mess of having different MSVC compilers
and
runtimes around. IMHO that part of Rashad’s issue is just a CMake bug
and not
a fundamental flaw in OSGeo4W.
Jürgen
–
Jürgen E. Fischer norBIT GmbH Tel.
+49-4931-918175-31
Dipl.-Inf. (FH) Rheinstraße 13 Fax.
+49-4931-918175-50
Software Engineer D-26506 Norden http://www.norbit.de
QGIS release manager (PSC) Germany IRC: jef on
FreeNode
