Skip to content

Commit 7c8d21b

Browse files
lsholsho
committed
Bug 36477854 - [36336179->14.1.2.0.0] Performance Degradation for UTF encoded strings After Upgrading .Net Client to 12.2.1.4, coherence-net-v14.1.1.0-core
[git-p4: depot-paths = "//dev/release.net/coherence-net-v14.1.1.0-core/": change = 108233]
1 parent 3faa7a2 commit 7c8d21b

File tree

7 files changed

+125
-330
lines changed

7 files changed

+125
-330
lines changed

‎src/Coherence.Core/IO/DataWriter.cs‎

Lines changed: 3 additions & 262 deletions
Original file line numberDiff line numberDiff line change
@@ -1,9 +1,10 @@
11
/*
2-
* Copyright (c) 2000, 2020, Oracle and/or its affiliates.
2+
* Copyright (c) 2000, 2024, Oracle and/or its affiliates.
33
*
44
* Licensed under the Universal Permissive License v 1.0 as shown at
55
* http://oss.oracle.com/licenses/upl.
66
*/
7+
78
using System;
89
using System.IO;
910
using System.Text;
@@ -48,31 +49,6 @@ public DataWriter(Stream output) : base(output)
4849

4950
#endregion
5051

51-
#region Properties
52-
53-
/// <summary>
54-
/// Obtain a temp buffer used to avoid allocations from
55-
/// repeated calls to String APIs.
56-
/// </summary>
57-
/// <return>
58-
/// a char buffer of CHAR_BUF_SIZE characters long
59-
/// </return>
60-
protected char[] CharBuf
61-
{
62-
get
63-
{
64-
// "partial" (i.e. windowed) char buffer just for formatUTF
65-
char[] ach = m_achBuf;
66-
if (ach == null)
67-
{
68-
m_achBuf = ach = new char[CHAR_BUF_SIZE];
69-
}
70-
return ach;
71-
}
72-
}
73-
74-
#endregion
75-
7652
#region Packed format writing
7753

7854
/// <summary>
@@ -403,247 +379,12 @@ public override void Write(string text)
403379
}
404380
else
405381
{
406-
byte[] bytes = FormatUTF(text);
382+
byte[] bytes = Encoding.UTF8.GetBytes(text);
407383
WritePackedInt32(bytes.Length);
408384
Write(bytes);
409385
}
410386
}
411387

412388
#endregion
413-
414-
#region UTF encoding functions
415-
416-
/// <summary>
417-
/// Figure out how many bytes it will take to hold the passed String.
418-
/// </summary>
419-
/// <remarks>
420-
/// This method is tightly bound to formatUTF.
421-
/// </remarks>
422-
/// <param name="s">
423-
/// the String
424-
/// </param>
425-
/// <return>
426-
/// the binary UTF length
427-
/// </return>
428-
protected int CalcUTF(String s)
429-
{
430-
int cch = s.Length;
431-
int cb = cch;
432-
char[] ach = CharBuf;
433-
bool fSmall = (cch <= CHAR_BUF_SIZE);
434-
if (fSmall)
435-
{
436-
var src = new StringBuilder(s);
437-
src.CopyTo(0, ach, 0, cch);
438-
}
439-
440-
for (int ofch = 0; ofch < cch; ++ofch)
441-
{
442-
int ch;
443-
if (fSmall)
444-
{
445-
ch = ach[ofch];
446-
}
447-
else
448-
{
449-
int ofBuf = ofch & CHAR_BUF_MASK;
450-
if (ofBuf == 0)
451-
{
452-
var src = new StringBuilder(s);
453-
int len = Math.Min(ofch + CHAR_BUF_SIZE, cch) - ofch;
454-
src.CopyTo(ofch, ach, 0, len);
455-
}
456-
ch = ach[ofBuf];
457-
}
458-
459-
if (ch <= 0x007F)
460-
{
461-
// all bytes in this range use the 1-byte format
462-
// except for 0
463-
if (ch == 0)
464-
{
465-
++cb;
466-
}
467-
}
468-
else
469-
{
470-
// either a 2-byte format or a 3-byte format (if over
471-
// 0x07FF)
472-
cb += (ch <= 0x07FF ? 1 : 2);
473-
}
474-
}
475-
476-
return cb;
477-
}
478-
479-
/// <summary>
480-
/// Format the passed String as UTF into the passed byte array.
481-
/// </summary>
482-
/// <remarks>
483-
/// This method is tightly bound to calcUTF.
484-
/// </remarks>
485-
/// <param name="s">
486-
/// the string.
487-
/// </param>
488-
/// <returns>
489-
/// The formated UTF byte array.
490-
/// </returns>
491-
public byte[] FormatUTF(String s)
492-
{
493-
int cch = s.Length;
494-
int cb = CalcUTF(s);
495-
int ofb = 0;
496-
byte[] ab = new byte[cb];
497-
498-
if (cb == cch)
499-
{
500-
// ask the string to convert itself to ascii bytes
501-
// straight into the WriteBuffer
502-
Encoding.ASCII.GetBytes(s, 0, cch, ab, ofb);
503-
}
504-
else
505-
{
506-
char[] ach = CharBuf;
507-
if (cch <= CHAR_BUF_SIZE)
508-
{
509-
// The following is unnecessary, because it would already
510-
// have been performed by calcUTF:
511-
//
512-
// if (fSmall)
513-
// {
514-
// s.getChars(0, cch, ach, 0);
515-
// }
516-
FormatUTF(ab, ofb, ach, cch);
517-
}
518-
else
519-
{
520-
for (int ofch = 0; ofch < cch; ofch += CHAR_BUF_SIZE)
521-
{
522-
int cchChunk = Math.Min(CHAR_BUF_SIZE, cch - ofch);
523-
StringBuilder src = new StringBuilder(s);
524-
src.CopyTo(ofch, ach, 0, cchChunk);
525-
ofb += FormatUTF(ab, ofb, ach, cchChunk);
526-
}
527-
}
528-
}
529-
530-
return ab;
531-
}
532-
533-
/// <summary>
534-
/// Format the passed characters as UTF into the passed byte array.
535-
/// </summary>
536-
/// <param name="ab">
537-
/// The byte array to format into.
538-
/// </param>
539-
/// <param name="ofb">
540-
/// The offset into the byte array to write the first byte.
541-
/// </param>
542-
/// <param name="ach">
543-
/// The array of characters to format.
544-
/// </param>
545-
/// <param name="cch">
546-
/// The number of characters to format.
547-
/// </param>
548-
/// <return>
549-
/// The number of bytes written to the array.
550-
/// </return>
551-
protected int FormatUTF(byte[] ab, int ofb, char[] ach, int cch)
552-
{
553-
int ofbOrig = ofb;
554-
for (int ofch = 0; ofch < cch; ++ofch)
555-
{
556-
char ch = ach[ofch];
557-
if (ch >= 0x0001 && ch <= 0x007F)
558-
{
559-
// 1-byte format: 0xxx xxxx
560-
ab[ofb++] = (byte) ch;
561-
}
562-
else if (ch <= 0x07FF)
563-
{
564-
// 2-byte format: 110x xxxx, 10xx xxxx
565-
ab[ofb++] = (byte) (0xC0 | ((ch >> 6) & 0x1F));
566-
ab[ofb++] = (byte) (0x80 | ((ch ) & 0x3F));
567-
}
568-
else
569-
{
570-
// 3-byte format: 1110 xxxx, 10xx xxxx, 10xx xxxx
571-
ab[ofb++] = (byte) (0xE0 | ((ch >> 12) & 0x0F));
572-
ab[ofb++] = (byte) (0x80 | ((ch >> 6) & 0x3F));
573-
ab[ofb++] = (byte) (0x80 | ((ch ) & 0x3F));
574-
}
575-
}
576-
return ofb - ofbOrig;
577-
}
578-
579-
///<summary>
580-
/// Get a buffer for formating data to bytes. Note that the resulting buffer
581-
/// may be shorter than the requested size.
582-
/// </summary>
583-
/// <param name="cb">
584-
/// the requested size for the buffer
585-
/// </param>
586-
/// <return>
587-
/// A byte array that is at least <tt>cb</tt> bytes long, but not
588-
/// shorter than <see cref="MIN_BUF"/> and (regardless of the value of
589-
/// <tt>cb</tt>) not longer than <see cref="MAX_BUF"/>.
590-
/// </return>
591-
protected byte[] Tmpbuf(int cb)
592-
{
593-
byte[] ab = m_abBuf;
594-
if (ab == null || ab.Length < cb)
595-
{
596-
int cbOld = ab == null ? 0 : ab.Length;
597-
int cbNew = Math.Max(MIN_BUF, Math.Min(MAX_BUF , cb));
598-
if (cbNew > cbOld)
599-
{
600-
m_abBuf = ab = new byte[cbNew > ((uint) MAX_BUF >> 1) ? MAX_BUF : cbNew];
601-
}
602-
}
603-
return ab;
604-
}
605-
606-
#endregion
607-
608-
#region Data Members
609-
610-
/// <summary>
611-
/// The minimum size of the temp buffer.
612-
/// </summary>
613-
private const int MIN_BUF = 0x40;
614-
615-
/// <summary>
616-
/// The maximum size of the temp buffer. The maximum size must be at least
617-
/// <tt>(3 * CHAR_BUF_SIZE)</tt> to accomodate the worst-case UTF
618-
/// formatting length.
619-
/// </summary>
620-
private const int MAX_BUF = 0x400;
621-
622-
/// <summary>
623-
/// Size of the temporary character buffer. Must be a power of 2.
624-
/// Size is: 256 characters (.25 KB).
625-
/// </summary>
626-
protected const int CHAR_BUF_SIZE = 0x100;
627-
628-
/// <summary>
629-
/// Bitmask used against a raw offset to determine the offset within
630-
/// the temporary character buffer.
631-
/// </summary>
632-
protected const int CHAR_BUF_MASK = (CHAR_BUF_SIZE - 1);
633-
634-
/// <summary>
635-
/// A temp buffer to use for building the data to write.
636-
/// </summary>
637-
[NonSerialized]
638-
private byte[] m_abBuf;
639-
640-
/// <summary>
641-
/// A lazily instantiated temp buffer used to avoid allocations from
642-
/// and repeated calls to String functions.
643-
/// </summary>
644-
[NonSerialized]
645-
protected char[] m_achBuf;
646-
647-
#endregion
648389
}
649390
}

‎src/Coherence.Core/IO/Pof/PofHelper.cs‎

Lines changed: 1 addition & 46 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
/*
2-
* Copyright (c) 2000, 2022, Oracle and/or its affiliates.
2+
* Copyright (c) 2000, 2024, Oracle and/or its affiliates.
33
*
44
* Licensed under the Universal Permissive License v 1.0 as shown at
55
* https://oss.oracle.com/licenses/upl.
@@ -313,51 +313,6 @@ public static int DecodeTinyInt(int n)
313313
/// </returns>
314314
public static char ReadChar(DataReader reader)
315315
{
316-
// int ch = reader.PeekChar();
317-
// if (ch == 65533) // Unicode replacement character
318-
// {
319-
// ch = reader.ReadByte();
320-
// int ch1 = ch & 0xFF;
321-
// switch ((ch1 & 0xF0) >> 4)
322-
// {
323-
// case 0xC:
324-
// case 0xD:
325-
// {
326-
// // 2-byte format: 110x xxxx, 10xx xxxx
327-
// int ch2 = reader.ReadByte() & 0xFF;
328-
// if ((ch2 & 0xC0) != 0x80)
329-
// {
330-
// throw new ArgumentException(
331-
// "illegal leading UTF byte: " + ch2);
332-
// }
333-
// ch = (char)(((ch1 & 0x1F) << 6) | ch2 & 0x3F);
334-
// break;
335-
// }
336-
//
337-
// case 0xE:
338-
// {
339-
// // 3-byte format: 1110 xxxx, 10xx xxxx, 10xx xxxx
340-
// int ch2 = reader.ReadByte() & 0xFF;
341-
// int ch3 = reader.ReadByte() & 0xFF;
342-
// if ((ch2 & 0xC0) != 0x80 || (ch3 & 0xC0) != 0x80)
343-
// {
344-
// throw new ArgumentException(
345-
// "illegal leading UTF bytes: " + ch2 + ", " + ch3);
346-
// }
347-
// ch = (char)(((ch & 0x0F) << 12) |
348-
// ((ch2 & 0x3F) << 6) |
349-
// ((ch3 & 0x3F)));
350-
// break;
351-
// }
352-
//
353-
// default:
354-
// throw new ArgumentException(
355-
// "illegal leading UTF byte: " + ch);
356-
// }
357-
//
358-
// return (char) ch;
359-
// }
360-
361316
return reader.ReadChar();
362317
}
363318

‎src/Coherence.Core/IO/Pof/WritingPofHandler.cs‎

Lines changed: 2 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -6,7 +6,7 @@
66
*/
77
using System;
88
using System.Diagnostics;
9-
using System.Text;
9+
1010
using Tangosol.Util;
1111

1212
namespace Tangosol.IO.Pof
@@ -744,14 +744,7 @@ public virtual void OnChar(int position, char ch)
744744
writer.WritePackedInt32(PofConstants.T_CHAR);
745745
}
746746

747-
if (Char.IsSurrogate(ch))
748-
{
749-
writer.Write(writer.FormatUTF(ch.ToString()));
750-
}
751-
else
752-
{
753-
writer.Write(ch);
754-
}
747+
writer.Write(ch);
755748
}
756749
}
757750

0 commit comments

Comments
 (0)