C＃R.NET错误：意外的'}'错误

Question

I'm using R.NET 1.6 on my program.

I got error message "Attempted to read or write protected memory. This is often an indication that other memory is corrupt" when I tried to run my whole script using source('') so I wrote the entire code manually instead of calling the script because I read somewhere that it's one of R.NET bug. Here's the code :

  engine.Evaluate("U=matrix(ncol=c,nrow=nrow(matdt))");
  engine.Evaluate("for(i in 1:nrow(matdt)){for(j in 1:c){U[i,j]=runif(1)}}");
  engine.Evaluate("totU=matrix(rowSums(U),ncol=1,nrow=nrow(U))");
  engine.Evaluate("for(a in 1:nrow(U)){for(b in 1:ncol(U)){U[a,b]=U[a,b]/totU[a,]}}");
  engine.Evaluate("P=0");
  engine.Evaluate("iter=1");
  engine.Evaluate("repeat{");
  engine.Evaluate("V=matrix(ncol=ncol(matdt),nrow=c)");
  engine.Evaluate("d=matrix(ncol=ncol(matdt),nrow=nrow(matdt))");                    
  engine.Evaluate("Uk=matrix(ncol=ncol(U),nrow=nrow(U))");
  engine.Evaluate("for(g in 1:nrow(U)){for(h in 1:ncol(U)){Uk[g,h]=(U[g,h])^(m)}}");                          engine.Evaluate("Uks=matrix(colSums(Uk),ncol=ncol(Uk),nrow=1)");
  engine.Evaluate("for(k in 1:c){for(j in 1:nrow(matdt)){d[j,]=Uk[j,k]*matdt[j,]}");
  engine.Evaluate("V[k,]=colSums(d)/Uks[,k]}");                                       
  engine.Evaluate("dist=matrix(ncol=1,nrow=nrow(matdt))");                             
  engine.Evaluate("alldist=matrix(ncol=c,nrow=nrow(matdt))");
  engine.Evaluate("for(l in 1:c){for(n in 1:nrow(matdt)){dist[n,]=t(matdt[n,]-V[l,])%*%(matdt[n,]-V[l,])}");
  engine.Evaluate("alldist[,l]=dist}");
  engine.Evaluate("ud=matrix(ncol=c,nrow=nrow(matdt))");
  engine.Evaluate("for(o in 1:nrow(matdt)){for(p in 1:c){ud[o,p]=alldist[o,p]*Uk[o,p]}}");
  engine.Evaluate("Pi=sum(rowSums(ud))");
  engine.Evaluate("P=abs(Pi-P)");
  engine.Evaluate("alldist2=matrix(ncol=ncol(alldist),nrow=nrow(alldist))");
  engine.Evaluate("for(q in 1:nrow(alldist)){for(r in 1:ncol(alldist)){alldist2[q,r]=(alldist[q,r])^((-1)/(m-1))}}");
  engine.Evaluate("alldist2[alldist2==Inf] <- 0 "); 
  engine.Evaluate("totdist=matrix(rowSums(alldist2),ncol=1,nrow=nrow(alldist2))");
  engine.Evaluate("for(s in 1:nrow(U)){for(t in 1:ncol(U)){U[s,t]=alldist2[s,t]/totdist[s,]}}");
  engine.Evaluate("nilnol<-which(U==0,arr.in=TRUE)");
  engine.Evaluate("if(nrow(nilnol)>0){nilnol=matrix(nilnol,ncol=ncol(nilnol))");
  engine.Evaluate("for(nn in 1:nrow(nilnol)){");                                              engine.Evaluate("U[nilnol[nn,1],]=0");                                                engine.Evaluate("U[nilnol[nn,1],nilnol[nn,2]]=1}}");
  engine.Evaluate("iter=iter+1");
  engine.Evaluate("if(P<e | iter>mit){break}}");

I got error message at engine.Evaluate("V[k,]=colSums(d)/Uks[,k]}"); saying unexpected '}' . It's the part of the loop :

engine.Evaluate("for(k in 1:c){for(j in 1:nrow(matdt)){d[j,]=Uk[j,k]*matdt[j,]}");
engine.Evaluate("V[k,]=colSums(d)/Uks[,k]}");

The code works fine when I tried to run it at R. Also if someone has solution for the bug problem it would be greatly appreciated.

Answer 1

Rather than numerous calls to engine.Evaluate() , you can construct longer strings of R commands and feed them into fewer calls. However, you'll need strategically placed semicolons; anywhere you would need a line break in regular R code you now need a semicolon.

In your case, you'll need a semicolon after the closing bracket of the for loop.

string cmd =
    "for (k in 1:c) {" +
        "for (j in 1:nrow(matdt)) {" +
            "d[j, ] = Uk[j, k]*matdt[j, ]" +
        "};" + /* <---- Note the semicolon */
        "V[k, ] = colSums(d) / Uks[, k]" +
    "};"; /* <---- Another just to be safe */

engine.Evaluate(cmd);

Since there are no embedded line breaks in the input string, the code that R is actually seeing is this:

for (k in 1:c) {for (j in 1:nrow(matdt)) {d[j, ] = Uk[j, k]*matdt[j, ]};V[k, ] = colSums(d) / Uks[, k]};

An easier solution that may also work is to pass a C# string literal that contains line breaks.

string cmd = @"for (k in 1:c) {
    for (j in 1:nrow(matdt)) {
        d[j, ] = Uk[k, j]*matdt[j, ]
    }
    V[k, ] = colSums(d) / Uks[, k]
}";

The line breaks are now embedded in the string, so when it's passed to engine.Evaluate() , R will see the line breaks as well, rendering semicolons unnecessary.