diff --git a/Numeric/GaussQuadraturePri.cpp b/Numeric/GaussQuadraturePri.cpp
index 7e6900479c81d153aebfe440a724ed5fd8c4bb5d..384db8d031d90f90136fd16c34c1cad1b576417e 100644
--- a/Numeric/GaussQuadraturePri.cpp
+++ b/Numeric/GaussQuadraturePri.cpp
@@ -3,54 +3,52 @@
 // See the LICENSE.txt file for license information. Please report all
 // bugs and problems to <gmsh@geuz.org>.
 
+#include "GmshMessage.h"
 #include "GaussIntegration.h"
 #include "GaussLegendre1D.h"
-#include "stdio.h"
 
 IntPt *getGQPriPts(int order);
 int getNGQPriPts(int order);
 
-IntPt * GQP[] = {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
+IntPt * GQP[] = {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+                 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
 
 IntPt *getGQPriPts(int order)
-{ 
+{
   int nLin = (order+3)/2;
   int nTri = getNGQTPts(order);
   int n = nLin*nTri;
   int index = order;
-  if (order >= sizeof(GQP) / sizeof(IntPt*))
+  if (order >= (int)(sizeof(GQP) / sizeof(IntPt*)))
     Msg::Fatal("Increase size of GQP in gauss quadrature prism");
-  if(!GQP[index])
-    {
-      double *linPt,*linWt;
-      IntPt *triPts = getGQTPts(order);
-//            printf("o = %d n = %d nT = %d nL = %d i = %d\n",order,n,nTri,nLin,index);
-      gmshGaussLegendre1D(nLin,&linPt,&linWt);
-      GQP[index] = new IntPt[n];
-      int l = 0;
-      for(int i=0; i < nTri; i++) {
-        for(int j=0; j < nLin; j++) {
-            GQP[index][l].pt[0] = triPts[i].pt[0];
-            GQP[index][l].pt[1] = triPts[i].pt[1];
-            GQP[index][l].pt[2] = linPt[j];
-            GQP[index][l++].weight = triPts[i].weight*linWt[j];
-//                  printf ("%d: %f %f %f %f\n",l-1,triPts[i].pt[0],triPts[i].pt[1],linPt[j],triPts[i].weight*linWt[j]);
-        }
+  if(!GQP[index]){
+    double *linPt,*linWt;
+    IntPt *triPts = getGQTPts(order);
+    gmshGaussLegendre1D(nLin,&linPt,&linWt);
+    GQP[index] = new IntPt[n];
+    int l = 0;
+    for(int i=0; i < nTri; i++) {
+      for(int j=0; j < nLin; j++) {
+        GQP[index][l].pt[0] = triPts[i].pt[0];
+        GQP[index][l].pt[1] = triPts[i].pt[1];
+        GQP[index][l].pt[2] = linPt[j];
+        GQP[index][l++].weight = triPts[i].weight*linWt[j];
       }
     }
+  }
   return GQP[index];
 }
 
 int getNGQPriPts(int order)
-{ 
+{
   int nLin = (order+3)/2;
   int nTri = getNGQTPts(order);
-  return nLin*nTri;
+  return nLin * nTri;
 
 //   if(order == 3)return 8;
 //   if(order == 2)return 8;
 //   if(order < 2)
-//     return GQPnPt[order]; 
+//     return GQPnPt[order];
 //   return ((order+3)/2)*((order+3)/2)*((order+3)/2);
 }
 
diff --git a/Numeric/fullMatrix.h b/Numeric/fullMatrix.h
index 0a13ed4cc88d77f20d2aa5611e44c1b7ea97b079..4741dc2c982423908f08ff2aebf21a110bbda079 100644
--- a/Numeric/fullMatrix.h
+++ b/Numeric/fullMatrix.h
@@ -541,8 +541,10 @@ class fullMatrix
     printf("};\n");
   }
 
-// specific functions for dgshell
-  void mult_naiveBlock(const fullMatrix<scalar> &b, const int ncol, const int fcol, const int alpha, const int beta, fullVector<scalar> &c, const int row=0) const
+  // specific functions for dgshell
+  void mult_naiveBlock(const fullMatrix<scalar> &b, const int ncol, const int fcol,
+                       const int alpha, const int beta, fullVector<scalar> &c,
+                       const int row=0) const
   {
     if(beta != 1)
       c.scale(beta);
@@ -550,7 +552,8 @@ class fullMatrix
       for(int k = 0; k < _c ; k++)
           c._data[j] += alpha*(*this)(row, k) * b(k, j);
   }
-  void multOnBlock(const fullMatrix<scalar> &b, const int ncol, const int fcol, const int alpha, const int beta, fullVector<scalar> &c) const
+  void multOnBlock(const fullMatrix<scalar> &b, const int ncol, const int fcol,
+                   const int alpha, const int beta, fullVector<scalar> &c) const
 #if !defined(HAVE_BLAS)
   {
     mult_naiveBlock(b,ncol,fcol,alpha,beta,c);
@@ -558,7 +561,8 @@ class fullMatrix
 #endif
   ;
 
-  void multWithATranspose(const fullVector<scalar> &x, scalar alpha, scalar beta, fullVector<scalar> &y) const
+  void multWithATranspose(const fullVector<scalar> &x, scalar alpha, scalar beta,
+                          fullVector<scalar> &y) const
 #if !defined(HAVE_BLAS)
   {
     y.scale(beta);
@@ -580,7 +584,8 @@ class fullMatrix
             scalar alpha=1., scalar beta=1.)
 #if !defined(HAVE_BLAS)
   {
-    Msg::Error("gemmWithAtranspose is only available with blas. If blas is not installed please transpose a before used gemm_naive");
+    Msg::Error("gemmWithAtranspose is only available with blas. If blas is not "
+               "installed please transpose a before used gemm_naive");
   }
 #endif
   ;
diff --git a/doc/gmsh.html b/doc/gmsh.html
index be84ebc610811add8ca3db220720a82179359ae9..cb11d9ee13d0a8b242bc7fea0d4e210d6909ec24 100644
--- a/doc/gmsh.html
+++ b/doc/gmsh.html
@@ -82,7 +82,8 @@ the <a href="/gmsh/doc/LICENSE.txt">GNU General Public License
   <li>Development version:
     <ul><li>automated nightly builds
         (<a href="http://onelab.info/CDash/index.php?project=Gmsh">dashboard</a>):
-        <a href="/gmsh/bin/Windows/gmsh-svn-Windows.zip">Windows</a>,
+        Windows <a href="/gmsh/bin/Windows/gmsh-svn-Windows.zip">32 bit</a> /
+                <a href="/gmsh/bin/Windows/gmsh-svn-Windows64.zip">64 bit</a>,
         Linux <a href="/gmsh/bin/Linux/gmsh-svn-Linux.tgz">32 bit</a> /
               <a href="/gmsh/bin/Linux/gmsh-svn-Linux64.tgz">64 bit</a>
          and