Thanks. I have modified and applied #1 and #2, which should make it into the next QL release. #3 will have to wait; GSLL is behind even later versions of GSL 1.x.
For patch #2, I have added the use of trivial-features to standardize the different platforms, so I use only #+windows now for any windows-based platform.
In the future, it would be helpful if you could not mix together unrelated changes into a single patch, and defer sending patches until you have figured out exactly where the problem is and how to fix it. For example, I did not apply the removal of basis-spline, because that works on other platforms, so this patch would gratuitously remove capability that functions for many people.
Thanks again, Liam
On Sun, Nov 22, 2015 at 4:47 PM, Mirko Vukovic mirko.vukovic@gmail.com wrote:
For GSLL I am attaching two git patches that accomplish several things:
- Loading of GSL that is part of MSYS2 on Windows
- Removed dependency on Osicat library
- Enable loading of GSL 2.0 by fixing a bit of code in nonlinear least
squares fitter.
- Disabled loading of basis-spline.lisp
- Cosmetic fixes.
I could not figure out the problem with basis-spline. I thus removed it. There may be a much better solution than that.
I checked the above on latest CCL and SBCL on Windows 7.
Warning: the two commits required some git surgery (splitting a commit into two). I hope that I did not introduce additional problems with the splits.
For Antik I am attaching one patch that fixes the readtable problem on CCL: (named-readtables:defreadtable :antik (:merge :standard) #-ccl(:macro-char ## :dispatch))
Without the patch all #... reader macros (CL and Antik) were deleted. With this patch both CL and Antik's reader macros work in CCL. I also tried the patch with SBCL, and I don't see a need for it. But since the macro-char redefinition did not hurt in SBCL, I left it in.
Mirko