diff --git a/client/gui_rpc_server_ops.cpp b/client/gui_rpc_server_ops.cpp
index 57009e2187..ba904a9ab1 100644
--- a/client/gui_rpc_server_ops.cpp
+++ b/client/gui_rpc_server_ops.cpp
@@ -908,43 +908,14 @@ static void handle_project_attach(GUI_RPC_CONN& grc) {
         }
     }
 
-	// remove http(s):// at the beginning of project address
-	// there is no reason to connect to secure address project
-	// if we're already connected to the non-secure address
-	// or vice versa
-	// also clear last '/' character if present
-
-    const string http = "http://";
-    const string https = "https://";
-
-    string new_project_url = url;
-    size_t pos = new_project_url.find(http);
-    if (pos != string::npos) {
-        new_project_url.erase(pos, http.length());
-    }
-    else if ((pos = new_project_url.find(https)) != string::npos) {
-        new_project_url.erase(pos, https.length());
-    }
-    if (new_project_url.length() >= 1 && new_project_url[new_project_url.length() - 1] == '/') {
-        new_project_url.erase(new_project_url.length() - 1, 1);
-	}
+    canonicalize_master_url(url);
 
     for (i=0; i<gstate.projects.size(); i++) {
         PROJECT* p = gstate.projects[i];
         string project_url = p->master_url;
+        canonicalize_master_url(project_url);
 
-        pos = project_url.find(http);
-        if (pos != string::npos) {
-            project_url.erase(pos, http.length());
-        }
-        else if ((pos = project_url.find(https)) != string::npos) {
-            project_url.erase(pos, https.length());
-        }
-        if (project_url.length() >= 1 && project_url[project_url.length() - 1] == '/') {
-            project_url.erase(project_url.length() - 1, 1);
-        }
-
-        if (new_project_url == project_url) {
+        if (url == project_url) {
             already_attached = true;
             break;
         }
diff --git a/lib/url.cpp b/lib/url.cpp
index 95bbb0e70d..ad7840eb12 100644
--- a/lib/url.cpp
+++ b/lib/url.cpp
@@ -203,6 +203,7 @@ void escape_url_readable(char *in, char* out) {
 // or prepend it
 //   - Remove double slashes in the rest
 //   - Add a trailing slash if necessary
+//   - Convert all alphabet characters to lower case
 //
 void canonicalize_master_url(char* url, int len) {
     char buf[1024];
@@ -225,6 +226,11 @@ void canonicalize_master_url(char* url, int len) {
     if (buf[n-1] != '/' && (n<sizeof(buf)-2)) {
         safe_strcat(buf, "/");
     }
+    for (size_t i=0; i<n-1; i++) {
+        // stop converting to lower-case, if we've reached the boundary of the domain name
+        if (buf[i] == '/') break;
+        buf[i] = tolower(static_cast<unsigned char>(buf[i]));
+    }
     snprintf(url, len, "http%s://%s", (bSSL ? "s" : ""), buf);
     url[len-1] = 0;
 }
diff --git a/tests/unit-tests/lib/test_url.cpp b/tests/unit-tests/lib/test_url.cpp
index 4e51a0b72e..71dd688acd 100644
--- a/tests/unit-tests/lib/test_url.cpp
+++ b/tests/unit-tests/lib/test_url.cpp
@@ -143,30 +143,56 @@ namespace test_url {
     }
 
     TEST_F(test_url, canonicalize_master_url) {
-        //Test to make sure a already good result comes back the same.
+        //Test to make sure an already good result comes back the same.
         string url = "http://secure.example.com/";
         canonicalize_master_url(url);
         EXPECT_STREQ(url.c_str(), "http://secure.example.com/");
 
-        //Test that https works, also adds trailing /
+        //Test to make sure an already good mixed-case result comes back lower-cased.
+        url = "http://SeCuRe.eXamPle.coM/";
+        canonicalize_master_url(url);
+        EXPECT_STREQ(url.c_str(), "http://secure.example.com/");
+
+        //Test that https works, also adds trailing /.
         url = "https://secure.example.com";
         canonicalize_master_url(url);
         EXPECT_STREQ(url.c_str(), "https://secure.example.com/");
 
-        //Test if someone forgot the leading http://
+        //Test that https works, in a mixed-case scenario, also adds trailing /.
+        url = "https://sEcUre.exaMple.cOm";
+        canonicalize_master_url(url);
+        EXPECT_STREQ(url.c_str(), "https://secure.example.com/");
+
+        //Test if someone forgot the leading http://.
         url = "www.example.com/";
         canonicalize_master_url(url);
         EXPECT_STREQ(url.c_str(), "http://www.example.com/");
 
-        //Test removing extra slashes.  And changing socks to https.
+        //Test omitted http:// and mixed case.
+        url = "wwW.exaMple.cOm/";
+        canonicalize_master_url(url);
+        EXPECT_STREQ(url.c_str(), "http://www.example.com/");
+
+        //Test removing extra slashes and changing socks to https.
         url = "socks://sock.example.com////////hello";
         canonicalize_master_url(url);
         EXPECT_STREQ(url.c_str(), "https://sock.example.com/hello/");
 
-        //Test if poorly written url
+        //Test removing extra slashes and changing socks to https, in a mixed-case scenario.
+        //Mixed-case characters after the domain name remain unaffected.
+        url = "sOcks://Sock.exaMPle.com////////hElLO";
+        canonicalize_master_url(url);
+        EXPECT_STREQ(url.c_str(), "https://sock.example.com/hElLO/");
+
+        //Test invalid protocol.
         url = "h://bad.example.com/";
         canonicalize_master_url(url);
         EXPECT_STREQ(url.c_str(), "http://bad.example.com/");
+
+        //Test invalid protocol and mixed case.
+        url = "H://baD.exampLE.Com/";
+        canonicalize_master_url(url);
+        EXPECT_STREQ(url.c_str(), "http://bad.example.com/");
     }
 
     TEST_F(test_url, valid_master_url) {