1 files changed, 28 insertions, 34 deletions

diff --git a/c++/lights_out.cpp b/c++/lights_out.cpp
index ae225d5..7dc8dc1 100644
--- a/c++/lights_out.cpp
+++ b/c++/lights_out.cpp
@@ -20,8 +20,6 @@ struct point {
     size_t x,y;
 
     point& inc() {
-        point& ret = *this;
-
         ++y;
         if(y >= NN){
             ++x;
@@ -32,7 +30,7 @@ struct point {
     }
 
     size_t stride() const {
-        return x + y*NN;
+        return x*NN + y;
     }
 
     bool equal(const point& rhs) const {
@@ -101,22 +99,16 @@ bool lu_decompose_gf2(Mat<MM,NN>& R, std::array<size_t,MM*NN>& P){
 }
 
 template<uint64_t MM, uint64_t NN>
-int solve_lights_out(std::array<int8_t,MM*NN>& b_raw){
-    Mat<M,N> A;
-		constexpr size_t MN = M*N;
-    
-    Vec<M,N> b{
-        1,0,1,
-        1,0,1,
-        0,0,0
-    };
+int solve_lights_out(const Vec<MM,NN>& b, Vec<MM,NN>& x){
+	Mat<MM,NN> A;
+	constexpr size_t MMNN = MM*NN;
 
     // Works best if we use odd N's
     // even N might not be solvable in roughly half of the cases
     // Init
     std::cout<<"Init"<<std::endl;
-    for(point<M,N> i{}; i.x < M && i.y < N; i.inc()){
-        for(point<M,N> j{}; j.x < M && j.y < N; j.inc()){
+    for(point<MM,NN> i{}; i.x < MM && i.y < NN; i.inc()){
+        for(point<MM,NN> j{}; j.x < MM && j.y < NN; j.inc()){
             if(
                 (i.equal(j)) ||
                 (i.x == j.x && (i.y+1) == j.y) ||
@@ -133,34 +125,29 @@ int solve_lights_out(std::array<int8_t,MM*NN>& b_raw){
         // Populate how you like it.
         //b[i.stride()] = false;
     }
-    std::array<size_t, MN> P_ident;
-    std::iota(P_ident.begin(),P_ident.end(),0);
-    print_mat<M,N>(A,P_ident);
-
-    std::array<size_t, MN> P;
+    std::array<size_t, MMNN> P;
     std::iota(P.begin(),P.end(),0);
     // Solve it :)
     std::cout<<"Solve"<<std::endl;
-    if(!lu_decompose_gf2<M,N>(A,P)){
+    if(!lu_decompose_gf2<MM,NN>(A,P)){
         return -1;
     }
-    std::array<size_t,MN> P_inv{};
-    for(size_t i = 0; i < MN; ++i){
+    std::array<size_t,MMNN> P_inv{};
+    for(size_t i = 0; i < MMNN; ++i){
         P_inv[P[i]] = i;
     }
 
-    Vec<M,N> y;
-    for(size_t i = 0; i < MN; ++i){
+    Vec<MM,NN> y;
+    for(size_t i = 0; i < MMNN; ++i){
         y[i] = b[P[i]];
         for(size_t j = 0; j < i; ++j){
             y[i] ^= A[P[i]][j] & y[j];
         }
     }
 
-    Vec<M,N> x;
-    for(size_t i = MN; i > 0; --i){
+    for(size_t i = MMNN; i > 0; --i){
         x[i-1] = y[i-1];
-        for(size_t j = i; j < MN; ++j){
+        for(size_t j = i; j < MMNN; ++j){
             x[i-1] ^= (A[P[i-1]][j] and x[j]);
         }
     }
@@ -169,14 +156,21 @@ int solve_lights_out(std::array<int8_t,MM*NN>& b_raw){
 }
 
 extern "C"{
-int solve_lights_out_3_x_3(std::array<int8_t, 3u*3u>& inp){
-	int rc = solve_lights_out<3u,3u>(inp);
-	return rc;
-}
+int solve_lights_out_dispatch(int8_t* b, int m, int n){
+	if ( m == 3 and n == 3){
+		Vec<3,3> b_raw;
+		Vec<3,3> x;
 
-int solve_lights_out_5_x_5(std::array<int8_t, 5u*5u>& inp){
-	int rc = solve_lights_out<5u,5u>(inp);
-	return rc;
+		auto rc = solve_lights_out<3,3>(b_raw,x);
+		if(rc == 0) {
+			for(point<3,3> i{}; i.x < 3 && i.y < 3; i.inc()){
+				auto is = i.stride();
+				b[is] = x[is];
+			}
+		}
+		return rc;
+	}
+	return -2;
 }
 
 int fake_main(){