Hi Clinton
I am one of those people who has packaged GSL for Windows. I did a 1.13
version, but I haven't had a burning need for a later version, so
haven't repackaged. But it's easy to do; I am happy to provide my build
scripts, too, if you want to try them and aren't allergic to MinGW, and
appropriately reference my efforts.
http://ascend4.org/Binary_installer_for_GSL-1.13_on_MinGW
One issue that I am aware of is 64-bit support. I haven't attempted to
build GSL on 64-bit, and that could be the source of the problems that
you refer to. Can I suggest that it would be extremely useful if you
were to spell out in more detail the types of problems that you have
experienced, and that will make it easier for problems to be
understood/fixed by the community.
Given that the C ABI is common across different compilers, there should
be no problem with MinGW DLLs and MSVC, but again, provided the
32-bit/64-bit issue is resolved.
Cheers
JP
On 05/12/11 09:56, Chee, Clinton wrote:
Dear GSL Development Team,
RE: Interest in Windows port of GSL
Thanks for your efforts in GSL - it is a very useful package.
I am writing to inform you of my Work In Progress in creating a Windows port.
My request and enquiry is when I finish the Windows port of GSL - would you be
interested in incorporating my changes into the main GSL code base.
Just to wet your taste-buds, I have put some notes of my progress at:
http://xtechnotes.blogspot.com/2011/11/notes-gsl.html
see the section called "Building using Visual Studio C/C++"
please note this is WORK IN PROGRESS, we can collaborate once I have finished
and am confident after all the testing.
At this stage I am just asking if you have future interest to assimilate my
work.
Reason for Windows PORT:
1. More reasons to have Windows GSL because Windows proprietary Stats library
are much more expensive. In the Linux world at least you can have open source
functions from various sources - of course none as comprehensive as GSL.
2. Windows GSL already exist for 1.8, 1.12 - Yes I found these. BUT they are
ported and made available as is. Their binaries may be available but they are
separate from the GSL development effort. So when GSL upgrades, these versions
are obsolete. More importantly these Windows port authors do not list how they
port their code, so future windows users cannot learn from them if they wish to
port future versions.
3. My port is using MSVC and maybe later Intel C/C++ for Windows. I tried
Windows-Linux emulators/environments like MinGW and have run into problems. I
have tried these for many days with lots of effort but I think MinGW is not
robust enough to handle GSL for Windows. In the Windows world, MSVC and Intel
C/C++ are quite common in scientific work and I am gearing towards these
platform.
Conclusion
Basically non of the Windows GSL option scattered across the world works for
me. And none of those people who ported to Windows seem to want to share their
techniques and instructions for porting to Windows. So I am doing my own port
of GSL using MSVC.
Once I finish it I am willing to share my work with the GSL Development team. I
am willing to show you what CHANGES I need to make to GSL to make it work. Most
importantly the CHANGES that I make is intended to be compatible with the
current LINUX-GNU/GSL. GSL would NOT need two separate code base for Windows
and Linux. My vision is if my changes are incorporated into the main GSL code,
then any user can download the source and compile simply using Microsoft Visual
Studio C/C++ and for Linux user, nothing would be changed.
Thanks for your consideration.
_______________________________________________
Help-gsl mailing list
Help-gsl@gnu.org
https://lists.gnu.org/mailman/listinfo/help-gsl
