Short-circuit Uri._uriEncode for components that don't need encoding.

sdk/lib/core/uri.dart

Index: sdk/lib/core/uri.dart
diff --git a/sdk/lib/core/uri.dart b/sdk/lib/core/uri.dart
index 10aa827cb190c6e93a8889cad67312629145d5f5..6955f5ccba6c3f2e1b3a6b2d2af9fada12f70ec9 100644
--- a/sdk/lib/core/uri.dart
+++ b/sdk/lib/core/uri.dart
@@ -2251,6 +2251,10 @@ class Uri {
   static const int _LOWER_CASE_Z = 0x7A;
   static const int _BAR = 0x7C;
 
+  // Matches a String that _uriEncodes to itself regardless of the kind of
+  // component.  This corresponds to [_unreservedTable].
+  static final RegExp _needsNoEncoding = new RegExp(r'^[\-\.0-9A-Z_a-z~]*$');

[Ivan Posva, 2015/11/04 05:11:50]

Please refrain from using RegExp in core library code. They are extremely
expensive to generate and should be avoided at all cost. Also they do not
necessarily execute fast depending on the execution platform or OS. Measuring on
a desktop gives you bad information in that matter.

+
   /**
    * This is the internal implementation of JavaScript's encodeURI function.
    * It encodes all characters in the string [text] except for those
@@ -2260,6 +2264,8 @@ class Uri {
                            String text,
                            {Encoding encoding: UTF8,
                             bool spaceToPlus: false}) {
+    if (_needsNoEncoding.hasMatch(text)) return text;

[Lasse Reichstein Nielsen, 2015/11/03 07:08:19]

This should be a regexp that matches canonicalTable. This function handles
encoding for all possible tables.

[Lasse Reichstein Nielsen, 2015/11/03 07:12:08]

Ah, finally realized that the only characters used are the ones that are not
encoded by any encoding table.

[Lasse Reichstein Nielsen, 2015/11/04 08:32:41]

Only do this if encoding is UTF8, LATIN1 or ASCII (or even only if it's UTF8
since that's the common case). 
Otherwise we don't know that an "A" is encoded as 0x41. I doubt we'll have
EBCDIC encoding (letters are above 128), but to be fair, we should allow for it.

It's too late to remove the encoding argument (if we want to reuse the function,
at least, there is only one call that actually passed the encoding).

If the optional parameters cause extra code, just make them non-optional. We can
call with ..., UTF8, false) in the places where the parameter is not passed.

+
     byteToHex(byte, buffer) {
       const String hex = '0123456789ABCDEF';
       buffer.writeCharCode(hex.codeUnitAt(byte >> 4));