Different output from toBase64() in CFML on 2 different machines - base64

Question

FINAL EDIT: SOLVED, upgrading local dev to railo 3.3.4.003 resolved the issue.
I have to RC4 encrypt some strings and have them base64 encoded and I'm running into a situation where the same input will generate different outputs on 2 different dev setups.
For instance, if I have a string test2#mail.com
On one machine (DEV-1) I'll get: DunU+ucIPz/Z7Ar+HTw=
and on the other (DEV-2) it'll be: DunU+ucIlZfZ7Ar+HTw=
First, I'm rc4 encrypting it through a function found here.
Next I'm feeding it to: toBase64( my_rc4_encrypted_data, "iso-8859-1")
As far as I can tell the rc4 encryption output is the same on both (or I'm missing something).
Below are SERVER variables from both machines as well as the encryption function.
Is this something we'll simply have to live with or is there something I can do to 'handle it properly' (for a lack of a better word).
I'm concerned that in the future this will bite me and wonder it it can be averted.
edit 1:
Output from my_rc4_encrypted_data.getBytes() returns:
dev-1:
Native Array (byte[])
14--23--44--6--25-8-63-63--39--20-10--2-29-60
dev-2:
Native Array (byte[])
14--23--44--6--25-8-63-63--39--20-10--2-29-60
(no encoding passed to getBytes() )
DEV-1 (remote)
server.coldfusion
productname Railo
productversion 9,0,0,1
server.java
archModel 64
vendor Sun Microsystems Inc.
version 1.6.0_26
server.os
arch amd64
archModel 64
name Windows Server 2008 R2
version 6.1
server.railo
version 3.3.2.002
server.servlet
name Resin/4.0.18
DEV-2 (local)
server.coldfusion
productname Railo
productversion 9,0,0,1
server.java
vendor Oracle Corporation
version 1.7.0_01
server.os
arch x86
name Windows 7
version 6.1
server.railo
version 3.2.2.000
server.servlet
name Resin/4.0.18
RC4 function:
function RC4(strPwd,plaintxt) {
var sbox = ArrayNew(1);
var key = ArrayNew(1);
var tempSwap = 0;
var a = 0;
var b = 0;
var intLength = len(strPwd);
var temp = 0;
var i = 0;
var j = 0;
var k = 0;
var cipherby = 0;
var cipher = "";
for(a=0; a lte 255; a=a+1) {
key[a + 1] = asc(mid(strPwd,(a MOD intLength)+1,1));
sbox[a + 1] = a;
}
for(a=0; a lte 255; a=a+1) {
b = (b + sbox[a + 1] + key[a + 1]) Mod 256;
tempSwap = sbox[a + 1];
sbox[a + 1] = sbox[b + 1];
sbox[b + 1] = tempSwap;
}
for(a=1; a lte len(plaintxt); a=a+1) {
i = (i + 1) mod 256;
j = (j + sbox[i + 1]) Mod 256;
temp = sbox[i + 1];
sbox[i + 1] = sbox[j + 1];
sbox[j + 1] = temp;
k = sbox[((sbox[i + 1] + sbox[j + 1]) mod 256) + 1];
cipherby = BitXor(asc(mid(plaintxt, a, 1)), k);
cipher = cipher & chr(cipherby);
}
return cipher;
}

Answer 1

Leigh wrote:
But be sure to use the same encoding in your test ie
String.getBytes(encoding) (Edit) If you omit it, the jvm default is
used.
Leigh is right - RAILO-1393 resulted in a change to toBase64 related to charset encodings in 3.3.0.017, which is between the 3.3.2.002 and 3.2.2.000 versions you are using.

Answer 2

As far as I can tell the rc4 encryption output is the same on both (or I'm missing something). Below are SERVER variables from both machines as well as the encryption function.
I would suggest saving the output to two files and then comparing the file size or, even better, a file comparison tool. Base64 encoding is a standard approach to converting binary data into string data.
Assuming that your binary files are both exactly 100% the same, on both of your servers try converting the data to base 64 and then back to binary again. I would predict that only one (or neither) of the servers are able to convert the data back to binary again. At that point, you should have a clue about which server is causing your problem and can dig in further.
If they both can reverse the base 64 data to binary and the binary is correct on both servers... well, I'm not sure.