[Gate-users] Gate on Windows

[Mojca Miklavec]

On Wed, Jun 26, 2013 at 10:41 PM, Marc Chamberland wrote:
> Searching the mailing list yields this result for Gate on Windows (credit to
> Marco Walterfang, apparently):
>
> http://article.gmane.org/gmane.comp.science.opengate.user/56/match=windows
>
> Note that this applies to Gate v5.0.
>
> Looking at the installation guide provided by Marco (see attached), there
> are a lot of small changes to be made to the source code. I'm not sure how
> much of this would apply to Gate v6.2.
>
> It could be a good starting point for anyone interested, though.

I didn't try to follow the PDF. (The PDF suggests installing MS Visual
Studio and cygwin. I bet that using cygwin can now be avoided, and
maybe even MSVS if really needed.)

Just for fun I tried compiling on Windows (using MinGW). I didn't come
as far as trying to install Gate, so there might be all kinds of new
problems with that one, but:

- CLHEP needed three trivial patches to compile (I submitted all of
the problems to the bug tracker and I hope that they will fix them)
and most probably compiles out-of-the-box with MSVC.

- ROOT failed already during CMake generation
(cmake/modules/SetUpWindows.cmake) when testing for the version of
visual studio. I would need some help compiling it with MinGW, but it
probably compiles with visual studio.

- Geant4 already failed with compilation at 2%. A bunch of fixes let
it proceed, however that is no guarantee that it would actually work.
I fixed some problems locally and reported them, but the last
compilation error was at 100% when it failed to find some OpenGL
functions and I wasn't sure how to fix that one. Again it seems that
Geant4 has good support for Visual Studio, but supporting MinGW should
be doable.

All in all it doesn't seem impossible to compile, but it makes more
sense to test with Microsoft's tools or one needs a bit of additional
work to be done, in particular with testing. If the Gate team didn't
use any unix-specific tricks, I believe that compiling Gate on Windows
is doable, even if not trivial for a newbie. ROOT and Geant4
admittedly use a lot of MSVS-specific code in cmake files, but let's
be optimists ...

It looks to me more or less the same as situation on Mac OS X some
time ago: no reason why it shouldn't work, but possibly tiny obstacles
standing on the way waiting for someone to fix them. And some
volunteer would be needed to provide precompiled binaries for Windows,
so that regular users wouldn't need to go through repeated lengthy
efforts to make it work.

I'm almost sure that it is a lot easier to build Gate 6.2 on Windows
now than it was for Gate 5.0 - before introduction of CMake-based
building system.

Mojca

> On 2013-06-26, at 4:27 PM, Mojca Miklavec wrote:
> On Wed, Jun 26, 2013 at 10:09 PM, Hermann Fuchs wrote:
>
>> Gate can be installed on Linx and MacOs.
>> As far as I know you can not install it on Windows.
>
>
> I'm not a windows user, but I'm just curious. Both Root and Geant4
> work fine on Windows and latest version of Gate uses CMake and
> supports Qt (no need for X11). Is there really so much
> platform-dependent code in Gate that makes it impossible to install on
> Windows? Are there some other dependencies of Gate that make
> particularly hard to compile it on windows (apart from the fact that
> setting up all dependencies for compiling Geant4 and Root is "hard
> enough")?
>
> Mojca