[logging-log4cxx] branch avoid_noop_encoding created (now 21461894)

[sw...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

swebb2066 pushed a change to branch avoid_noop_encoding
in repository https://gitbox.apache.org/repos/asf/logging-log4cxx.git


      at 21461894 Reduce logging overhead when the output is unchanged by the character encoding

This branch includes the following new commits:

     new 21461894 Reduce logging overhead when the output is unchanged by the character encoding

The 1 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails.  The revisions
listed as "add" were already present in the repository and have only
been added to this reference.

[logging-log4cxx] 01/01: Reduce logging overhead when the output is unchanged by the character encoding

[sw...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

swebb2066 pushed a commit to branch avoid_noop_encoding
in repository https://gitbox.apache.org/repos/asf/logging-log4cxx.git

commit 214618948d82ff3bcc6f73b87a77d0a4b5a7c28c
Author: Stephen Webb <sw...@gmail.com>
AuthorDate: Wed Aug 9 16:10:42 2023 +1000

    Reduce logging overhead when the output is unchanged by the character encoding
---
 src/main/cpp/charsetencoder.cpp                   | 15 ++++++++++++
 src/main/cpp/outputstreamwriter.cpp               | 30 ++++++++++++++---------
 src/main/include/log4cxx/helpers/charsetencoder.h |  6 +++++
 3 files changed, 40 insertions(+), 11 deletions(-)

diff --git a/src/main/cpp/charsetencoder.cpp b/src/main/cpp/charsetencoder.cpp
index df61e487..5f08b103 100644
--- a/src/main/cpp/charsetencoder.cpp
+++ b/src/main/cpp/charsetencoder.cpp
@@ -644,3 +644,18 @@ void CharsetEncoder::encode(CharsetEncoderPtr& enc,
 		dst.put(Transcoder::LOSSCHAR);
 	}
 }
+
+bool CharsetEncoder::isTriviallyCopyable(const LogString& src, CharsetEncoderPtr& enc)
+{
+	bool result = false;
+	if (dynamic_cast<LocaleCharsetEncoder*>(enc.get()))
+	{
+		result = src.end() == std::find_if(src.begin(), src.end()
+			, [](const logchar& ch) -> bool { return 0x80 <= (unsigned int)ch; });
+	}
+#if LOG4CXX_LOGCHAR_IS_UTF8
+	else
+		result = !!dynamic_cast<TrivialCharsetEncoder*>(enc.get());
+#endif
+	return result;
+}
diff --git a/src/main/cpp/outputstreamwriter.cpp b/src/main/cpp/outputstreamwriter.cpp
index 4ddf93c9..1ebd320e 100644
--- a/src/main/cpp/outputstreamwriter.cpp
+++ b/src/main/cpp/outputstreamwriter.cpp
@@ -78,18 +78,29 @@ void OutputStreamWriter::flush(Pool& p)
 
 void OutputStreamWriter::write(const LogString& str, Pool& p)
 {
-	if (str.length() > 0)
+	if (str.empty())
+		return;
+	if (CharsetEncoder::isTriviallyCopyable(str, m_priv->enc))
 	{
+		ByteBuffer buf((char*)str.data(), str.size() * sizeof (logchar));
+		m_priv->out->write(buf, p);
+	}
+	else
+	{
+		enum { BUFSIZE = 1024 };
+		char stackData[BUFSIZE];
+		char* rawbuf = stackData;
+		size_t bufSize = BUFSIZE;
 #ifdef LOG4CXX_MULTI_PROCESS
+		std::vector<char> heapData;
 		// Ensure the logging event is a single write system call to keep events from each process separate
-		size_t bufSize = str.length() * 2;
-		char* rawbuf = new char[bufSize];
-		ByteBuffer buf(rawbuf, (size_t) bufSize);
-#else
-		enum { BUFSIZE = 1024 };
-		char rawbuf[BUFSIZE];
-		ByteBuffer buf(rawbuf, (size_t) BUFSIZE);
+		if (bufSize < str.length() * 2)
+		{
+			heapData.resize(bufSize = str.length() * 2)
+			rawbuf = heapData.data();
+		}
 #endif
+		ByteBuffer buf(rawbuf, bufSize);
 		m_priv->enc->reset();
 		LogString::const_iterator iter = str.begin();
 
@@ -105,9 +116,6 @@ void OutputStreamWriter::write(const LogString& str, Pool& p)
 		m_priv->enc->flush(buf);
 		buf.flip();
 		m_priv->out->write(buf, p);
-#ifdef LOG4CXX_MULTI_PROCESS
-		delete []rawbuf;
-#endif
 	}
 }
 
diff --git a/src/main/include/log4cxx/helpers/charsetencoder.h b/src/main/include/log4cxx/helpers/charsetencoder.h
index c54e35dd..f85d8451 100644
--- a/src/main/include/log4cxx/helpers/charsetencoder.h
+++ b/src/main/include/log4cxx/helpers/charsetencoder.h
@@ -116,6 +116,12 @@ class LOG4CXX_EXPORT CharsetEncoder : public Object
 			return (stat != 0);
 		}
 
+		/**
+		* Is the data of \c src unchanged by \c enc.
+		*
+		*/
+		static bool isTriviallyCopyable(const LogString& src, CharsetEncoderPtr& enc);
+
 
 	private:
 		/**