无论是Android亦或者Java中或多或少需要调用底层的一些命令,执行一些参数;

此时我们需要用到Java的Process来创建一个子进程,之所以是子进程是因为此进程依赖于发起创建请求的进程,如果发起者被Kill那个子进程也将Kill。

对于Process相信使用过的朋友一定不会陌生,它具有如下特点:

1.创建简单

2.控制难

3.容易导致无法创建子进程

4.如果是多线程那么很有可能造成内存溢出

以上现象如果你只是偶尔使用一次,创建一个进程或许你什么都没有感觉到,但是如果你使用了多线程,进行了大量的创建,以上问题你都会遇到。

相关:[Android] ProcessBuilder与Runtime.getRuntime().exec分别创建进程的区别,[Android] [Java] 分享 Process 执行命令行封装类

这两个星期一直在研究上面的问题,我要做的软件是在Android中进行TraceRoute,由于手机不可能完全Root所以不能采用JNI来发送ICMP请求的方式,最终只能使用创建进程方式进行;具体实现思路是:使用PING命令来PING百度等地址,在PING命令中加入TTL,得到每一次的IP地址,当IP地址与目标IP地址符合时退出,并且还需要单独PING一次每一跳的延迟和丢包。

单线程:PING 百度 TTL=1 =》 得到IP,PING IP 得到延迟丢包,改变TTL,进行下一次PING,直到所得到的IP与目标(百度)一样时停止。按照上面的思路一次需要创建两个子进程,一般到百度时TTL大约为12跳左右,所以就是2*12=24个子进程;如果是在单线程下简单明了,但是速度慢,整个过程大约需要1分钟左右。

多线程:同时发起3个线程进行3跳测试TTL=(1,2,3),测试完成后测试下一批数据TTL=(4,5,6),如果也是12跳的话,那么也是24个子进程,但是整体耗时将会为1/3.可见此时效率较高。

但是多线程需要考虑的是线程的同步问题,以及得到数据后的写入问题,这些赞不谈,只谈进程问题。经过我的测试假如现在测试100个网站的TraceRoute数据,在上层控制一次测试4个网站,底层实现并发3个线程,此时在一定时间内将会同时存在3*4个进程。按照平均每个网站12跳来算:12*2*100=240个子进程,需要的子线程为12*100=120个。

这个时候问题来了,假如现在程序子进程不正常了,遇到了一个一定的问题导致进程无法执行完成,此时你的现象是:一个子进程卡住,随后创建的所有子进程都卡住。假如最上层线程做了任务时间限制,那么到时间后将会尝试销毁,但是你会发现无法销毁,所持有的线程也不会销毁。但是上层以为销毁掉了,然后继续进行下一批的数据测试,此时你的线程数量会逐渐增加,如果100任务下来你的线程或许会达到3*4*100=1200如果有前期没有这样的情况那个就是一半:600个线程左右,如果后期还有任务将会继续增加但是却永远不会销毁,但是我们知道JVM的内存是有限的,所以此时将会出现内存溢出。

以上就是我遇到的问题,我最先改为了等待线程完全返回后再进行下一批数据测试,此时内存溢出是解决了,但是任务却一直卡住在哪里了,永远也不走。我就在想要解决这一的问题需要解决根本上的问题才行,经过研究我发现在程序创建了子进程后JVM将会创建一个子进程管理线程:“ProcessManager”:

正常情况下该线程状态为Native,但是如果创建大量子进程后有可能会出现此线程为Monitor状态,过一段时间后所有创建子进程的线程状态也将会变为Monitor状态,然后将一直死锁,后面创建线程也是继续死锁,无法继续。

通过查看ProcessManager源码发现,其中启动了一个线程用于监听子进程状态,同时管理子进程,比如输出消息以及关闭子进程等操作,具体如下:

/*** Copyright (C) 2007 The Android Open Source Project** Licensed under the Apache License, Version 2.0 (the "License");* you may not use this file except in compliance with the License.* You may obtain a copy of the License at**      http://www.apache.org/licenses/LICENSE-2.0** Unless required by applicable law or agreed to in writing, software* distributed under the License is distributed on an "AS IS" BASIS,* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.* See the License for the specific language governing permissions and* limitations under the License.*/package java.lang;import java.io.File;
import java.io.FileDescriptor;
import java.io.FileInputStream;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.InputStream;
import java.io.OutputStream;
import java.lang.ref.ReferenceQueue;
import java.lang.ref.WeakReference;
import java.util.HashMap;
import java.util.Map;
import java.util.Arrays;
import java.util.logging.Logger;
import java.util.logging.Level;/**** Manages child processes.** <p>Harmony's native implementation (for comparison purposes):* http://tinyurl.com/3ytwuq*/
final class ProcessManager {/**** constant communicated from native code indicating that a* child died, but it was unable to determine the status*/private static final int WAIT_STATUS_UNKNOWN = -1;/**** constant communicated from native code indicating that there* are currently no children to wait for*/private static final int WAIT_STATUS_NO_CHILDREN = -2;/**** constant communicated from native code indicating that a wait()* call returned -1 and set an undocumented (and hence unexpected) errno*/private static final int WAIT_STATUS_STRANGE_ERRNO = -3;/**** Initializes native static state.*/static native void staticInitialize();static {staticInitialize();}/**** Map from pid to Process. We keep weak references to the Process objects* and clean up the entries when no more external references are left. The* process objects themselves don't require much memory, but file* descriptors (associated with stdin/out/err in this case) can be* a scarce resource.*/private final Map<Integer, ProcessReference> processReferences= new HashMap<Integer, ProcessReference>();/*** Keeps track of garbage-collected Processes. */private final ProcessReferenceQueue referenceQueue= new ProcessReferenceQueue();private ProcessManager() {// Spawn a thread to listen for signals from child processes.Thread processThread = new Thread(ProcessManager.class.getName()) {@Overridepublic void run() {watchChildren();}};processThread.setDaemon(true);processThread.start();}/**** Kills the process with the given ID.** @parm pid ID of process to kill*/private static native void kill(int pid) throws IOException;/**** Cleans up after garbage collected processes. Requires the lock on the* map.*/void cleanUp() {ProcessReference reference;while ((reference = referenceQueue.poll()) != null) {synchronized (processReferences) {processReferences.remove(reference.processId);}}}/**** Listens for signals from processes and calls back to* {@link #onExit(int,int)}.*/native void watchChildren();/**** Called by {@link #watchChildren()} when a child process exits.** @param pid ID of process that exited* @param exitValue value the process returned upon exit*/void onExit(int pid, int exitValue) {ProcessReference processReference = null;synchronized (processReferences) {cleanUp();if (pid >= 0) {processReference = processReferences.remove(pid);} else if (exitValue == WAIT_STATUS_NO_CHILDREN) {if (processReferences.isEmpty()) {/*** There are no eligible children; wait for one to be* added. The wait() will return due to the* notifyAll() call below.*/try {processReferences.wait();} catch (InterruptedException ex) {// This should never happen.throw new AssertionError("unexpected interrupt");}} else {/*** A new child was spawned just before we entered* the synchronized block. We can just fall through* without doing anything special and land back in* the native wait().*/}} else {// Something weird is happening; abort!throw new AssertionError("unexpected wait() behavior");}}if (processReference != null) {ProcessImpl process = processReference.get();if (process != null) {process.setExitValue(exitValue);}}}/**** Executes a native process. Fills in in, out, and err and returns the* new process ID upon success.*/static native int exec(String[] command, String[] environment,String workingDirectory, FileDescriptor in, FileDescriptor out,FileDescriptor err, boolean redirectErrorStream) throws IOException;/**** Executes a process and returns an object representing it.*/Process exec(String[] taintedCommand, String[] taintedEnvironment, File workingDirectory,boolean redirectErrorStream) throws IOException {// Make sure we throw the same exceptions as the RI.if (taintedCommand == null) {throw new NullPointerException();}if (taintedCommand.length == 0) {throw new IndexOutOfBoundsException();}// Handle security and safety by copying mutable inputs and checking them.String[] command = taintedCommand.clone();String[] environment = taintedEnvironment != null ? taintedEnvironment.clone() : null;SecurityManager securityManager = System.getSecurityManager();if (securityManager != null) {securityManager.checkExec(command[0]);}// Check we're not passing null Strings to the native exec.for (String arg : command) {if (arg == null) {throw new NullPointerException();}}// The environment is allowed to be null or empty, but no element may be null.if (environment != null) {for (String env : environment) {if (env == null) {throw new NullPointerException();}}}FileDescriptor in = new FileDescriptor();FileDescriptor out = new FileDescriptor();FileDescriptor err = new FileDescriptor();String workingPath = (workingDirectory == null)? null: workingDirectory.getPath();// Ensure onExit() doesn't access the process map before we add our// entry.synchronized (processReferences) {int pid;try {pid = exec(command, environment, workingPath, in, out, err, redirectErrorStream);} catch (IOException e) {IOException wrapper = new IOException("Error running exec()." + " Command: " + Arrays.toString(command)+ " Working Directory: " + workingDirectory+ " Environment: " + Arrays.toString(environment));wrapper.initCause(e);throw wrapper;}ProcessImpl process = new ProcessImpl(pid, in, out, err);ProcessReference processReference= new ProcessReference(process, referenceQueue);processReferences.put(pid, processReference);/*** This will wake up the child monitor thread in case there* weren't previously any children to wait on.*/processReferences.notifyAll();return process;}}static class ProcessImpl extends Process {/*** Process ID. */final int id;final InputStream errorStream;/*** Reads output from process. */final InputStream inputStream;/*** Sends output to process. */final OutputStream outputStream;/*** The process's exit value. */Integer exitValue = null;final Object exitValueMutex = new Object();ProcessImpl(int id, FileDescriptor in, FileDescriptor out,FileDescriptor err) {this.id = id;this.errorStream = new ProcessInputStream(err);this.inputStream = new ProcessInputStream(in);this.outputStream = new ProcessOutputStream(out);}public void destroy() {try {kill(this.id);} catch (IOException e) {Logger.getLogger(Runtime.class.getName()).log(Level.FINE,"Failed to destroy process " + id + ".", e);}}public int exitValue() {synchronized (exitValueMutex) {if (exitValue == null) {throw new IllegalThreadStateException("Process has not yet terminated.");}return exitValue;}}public InputStream getErrorStream() {return this.errorStream;}public InputStream getInputStream() {return this.inputStream;}public OutputStream getOutputStream() {return this.outputStream;}public int waitFor() throws InterruptedException {synchronized (exitValueMutex) {while (exitValue == null) {exitValueMutex.wait();}return exitValue;}}void setExitValue(int exitValue) {synchronized (exitValueMutex) {this.exitValue = exitValue;exitValueMutex.notifyAll();}}@Overridepublic String toString() {return "Process[id=" + id + "]";  }}static class ProcessReference extends WeakReference<ProcessImpl> {final int processId;public ProcessReference(ProcessImpl referent,ProcessReferenceQueue referenceQueue) {super(referent, referenceQueue);this.processId = referent.id;}}static class ProcessReferenceQueue extends ReferenceQueue<ProcessImpl> {@Overridepublic ProcessReference poll() {// Why couldn't they get the generics right on ReferenceQueue? :(Object reference = super.poll();return (ProcessReference) reference;}}static final ProcessManager instance = new ProcessManager();/*** Gets the process manager. */static ProcessManager getInstance() {return instance;}/*** Automatically closes fd when collected. */private static class ProcessInputStream extends FileInputStream {private FileDescriptor fd;private ProcessInputStream(FileDescriptor fd) {super(fd);this.fd = fd;}@Overridepublic void close() throws IOException {try {super.close();} finally {synchronized (this) {if (fd != null && fd.valid()) {try {ProcessManager.close(fd);} finally {fd = null;}}}}}}/*** Automatically closes fd when collected. */private static class ProcessOutputStream extends FileOutputStream {private FileDescriptor fd;private ProcessOutputStream(FileDescriptor fd) {super(fd);this.fd = fd;}@Overridepublic void close() throws IOException {try {super.close();} finally {synchronized (this) {if (fd != null && fd.valid()) {try {ProcessManager.close(fd);} finally {fd = null;}}}}}}/*** Closes the given file descriptor. */private static native void close(FileDescriptor fd) throws IOException;
}

在其中有一个“ native void watchChildren();”方法,此方法为线程主方法,具体实现可以看看JNI,在其中回调了方法:“ void onExit(int pid, int exitValue);” 在方法中:

void onExit(int pid, int exitValue) {ProcessReference processReference = null;synchronized (processReferences) {cleanUp();if (pid >= 0) {processReference = processReferences.remove(pid);} else if (exitValue == WAIT_STATUS_NO_CHILDREN) {if (processReferences.isEmpty()) {/*** There are no eligible children; wait for one to be* added. The wait() will return due to the* notifyAll() call below.*/try {processReferences.wait();} catch (InterruptedException ex) {// This should never happen.throw new AssertionError("unexpected interrupt");}} else {/*** A new child was spawned just before we entered* the synchronized block. We can just fall through* without doing anything special and land back in* the native wait().*/}} else {// Something weird is happening; abort!throw new AssertionError("unexpected wait() behavior");}}if (processReference != null) {ProcessImpl process = processReference.get();if (process != null) {process.setExitValue(exitValue);}}}

此方法作用是删除子进程队列中子进程同时通知子进程 ProcessImpl已完成。

但是在方法:“watchChildren()”中如果出现System.in缓冲期满的情况那么进程将无法正常结束,它将一直等待缓冲区有空间存在,而缓冲区又是公共区间,如果一个出现等待那么后续子进程也将全部等待,如果缓冲区无法清空,那么所有子进程将会全部死锁掉。这就是导致子进程卡死的凶手。

知道问题关键点那么就会有人想办法解决,例如:

//...读取数据...process.waitFor();//....再次读取

这样的方式看似很好,但是你有没有想过有些数据无法及时返回,所以在 waitfor()之前读取很有可能没有数据导致进行 waitfor()等待,这时我们可以看看源码:

        public int waitFor() throws InterruptedException {synchronized (exitValueMutex) {while (exitValue == null) {exitValueMutex.wait();}return exitValue;}}
        void setExitValue(int exitValue) {synchronized (exitValueMutex) {this.exitValue = exitValue;exitValueMutex.notifyAll();}}

这里可以看见假如没有退出值将会进行等待,直到通知发生,但是通知想要发生必须要靠“ ProcessManager ”线程来告诉你。但是假如在等待过程中出现了大量的数据,导致 System.IN 满了,此时“ ProcessManager ”线程很傻很傻的进入了等待状态中,也将无法进行通知,而这边也就无法往下走,无法到达第二次读取,所以第二次读取就很随机了,在大量数据下第二次读取基本上就是摆设,也就是说无法正常的执行,最终也将导致死锁。

解决办法也很简单,创建线程后我们可以创建一个线程来专门读取信息,直到“ProcessManager”线程通知结束的时候,才退出线程。

首先我们看看Process提供的“exitValue()”方法:

        public int exitValue() {synchronized (exitValueMutex) {if (exitValue == null) {throw new IllegalThreadStateException("Process has not yet terminated.");}return exitValue;}}

可见在” exitValue “没有值时将会抛出异常而不会阻塞,所以可以得出:” exitValue() “与” waitfor() “都可以用于判断线程是否完成,但是一个是阻塞的一个是不阻塞的方法,在线程中当然使用不阻塞的来完成我们的工作:

    /*** 实例化一个ProcessModel** @param process Process*/private ProcessModel(Process process) {//initthis.process = process;//getout = process.getOutputStream();in = process.getInputStream();err = process.getErrorStream();//inif (in != null) {isInReader = new InputStreamReader(in);bInReader = new BufferedReader(isInReader, BUFFER_LENGTH);}sbReader = new StringBuilder();//start read threadreadThread();}....................//读取结果private void read() {String str;//read Intry {while ((str = bInReader.readLine()) != null) {sbReader.append(str);sbReader.append(BREAK_LINE);}} catch (Exception e) {e.printStackTrace();Logs.e(TAG, e.getMessage());}}/*** 启动线程进行异步读取结果*/private void readThread() {Thread thread = new Thread(new Runnable() {@Overridepublic void run() {//while (true) {try {process.exitValue();//read lastread();break;} catch (IllegalThreadStateException e) {read();}StaticFunction.sleepIgnoreInterrupt(300);}//read endint len;if (in != null) {try {while ((len = in.read(BUFFER)) > 0) {Logs.d(TAG, String.valueOf(len));}} catch (IOException e) {e.printStackTrace();Logs.e(TAG, e.getMessage());}}//closeclose();//doneisDone = true;}});thread.setName("DroidTestAgent.Test.TestModel.ProcessModel:ReadThread");thread.setDaemon(true);thread.start();}

当创建进程后把进程丢进我建立的类中实例化为一个进程管理类,随后启动线程,线程执行中调用进程的” exitValue()“ ,如果异常就进入读取数据,直到不异常时再次读取一次最后数据,随后退出循环,退出后还读取了一次底层的数据(这个其实可以不用要,纯属心理作用!)。最后写入完成标记。其中” StaticFunction.sleepIgnoreInterrupt(300); “是我写的静态方法用于休眠等待而已,也就是 Sleep ,只不过加入了 try catch 。

当然光是读取IN流是不行的,还有Error流,这个时候就需要两个线程来完成,一个也行。不过我为了简单采用了:ProcessBuilder类创建进程并重定向了错误流到IN流中,这样简化了操作。

而使用ProcessBuilder类需要注意的是同一个ProcessBuilder实例创建子进程的时候是需要进行线程同步操作的,因为如果并发操作将会导致进程参数错误等现象发生,所以建议加上线程互斥来实现,但是不建议重复创建ProcessBuilder实例,创建那么多实例,何不把所有子进程放在一个ProcessBuilder实例里边。减少内存消耗啊,手机伤不起啊。

有必要提出的是,当线程判断结束的时候,也就是退出值(exitvalue)有值得时候此时其实在”ProcessManager“线程中已经杀掉了进程了,此时在进程中其实没有此进程了,有的也就是执行后的数据流而已。所以正常结束情况下无需自己调用”destroy()“方法,调用后将会触发异常,说没有找到此进程。

        public void destroy() {try {kill(this.id);} catch (IOException e) {Logger.getLogger(Runtime.class.getName()).log(Level.FINE,"Failed to destroy process " + id + ".", e);}}

终于讲完了,累啊;

最后给大家分享我自己弄得一个类(ProcessModel),大家喜欢就直接拿去,如果有好的建议希望大家提出来:

import com.droidtestagent.journal.Logs;
import com.droidtestagent.util.StaticFunction;import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.io.OutputStream;
import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;/*** Create By Qiujuer* 2014-08-05* <p/>* 执行命令行语句进程管理封装*/
public class ProcessModel {private static final String TAG = "ProcessModel";//换行符private static final String BREAK_LINE;//错误缓冲private static final byte[] BUFFER;//缓冲区大小private static final int BUFFER_LENGTH;//创建进程时需要互斥进行private static final Lock lock = new ReentrantLock();//ProcessBuilderprivate static final ProcessBuilder prc;final private Process process;final private InputStream in;final private InputStream err;final private OutputStream out;final private StringBuilder sbReader;private BufferedReader bInReader = null;private InputStreamReader isInReader = null;private boolean isDone;/*** 静态变量初始化*/static {BREAK_LINE = "\n";BUFFER_LENGTH = 128;BUFFER = new byte[BUFFER_LENGTH];prc = new ProcessBuilder();}/*** 实例化一个ProcessModel** @param process Process*/private ProcessModel(Process process) {//initthis.process = process;//getout = process.getOutputStream();in = process.getInputStream();err = process.getErrorStream();//inif (in != null) {isInReader = new InputStreamReader(in);bInReader = new BufferedReader(isInReader, BUFFER_LENGTH);}sbReader = new StringBuilder();//start read threadreadThread();}/*** 执行命令** @param params 命令参数 eg: "/system/bin/ping", "-c", "4", "-s", "100","www.qiujuer.net"*/public static ProcessModel create(String... params) {Process process = null;try {lock.lock();process = prc.command(params).redirectErrorStream(true).start();} catch (IOException e) {e.printStackTrace();} finally {//sleep 100StaticFunction.sleepIgnoreInterrupt(100);lock.unlock();}if (process == null)return null;return new ProcessModel(process);}/*** 通过Android底层实现进程关闭** @param process 进程*/public static void kill(Process process) {int pid = getProcessId(process);if (pid != 0) {try {android.os.Process.killProcess(pid);} catch (Exception e) {try {process.destroy();} catch (Exception ex) {//ex.printStackTrace();}}}}/*** 获取进程的ID** @param process 进程* @return id*/public static int getProcessId(Process process) {String str = process.toString();try {int i = str.indexOf("=") + 1;int j = str.indexOf("]");str = str.substring(i, j);return Integer.parseInt(str);} catch (Exception e) {return 0;}}//读取结果private void read() {String str;//read Intry {while ((str = bInReader.readLine()) != null) {sbReader.append(str);sbReader.append(BREAK_LINE);}} catch (Exception e) {e.printStackTrace();Logs.e(TAG, e.getMessage());}}/*** 启动线程进行异步读取结果*/private void readThread() {Thread thread = new Thread(new Runnable() {@Overridepublic void run() {//while to endwhile (true) {try {process.exitValue();//read lastread();break;} catch (IllegalThreadStateException e) {read();}StaticFunction.sleepIgnoreInterrupt(300);}//read endint len;if (in != null) {try {while ((len = in.read(BUFFER)) > 0) {Logs.d(TAG, String.valueOf(len));}} catch (IOException e) {e.printStackTrace();Logs.e(TAG, e.getMessage());}}//closeclose();//doneisDone = true;}});thread.setName("DroidTestAgent.Test.TestModel.ProcessModel:ReadThread");thread.setDaemon(true);thread.start();}/*** 获取执行结果** @return 结果*/public String getResult() {//waite process setValuetry {process.waitFor();} catch (Exception e) {e.printStackTrace();Logs.e(TAG, e.getMessage());}//until startRead enwhile (true) {if (isDone)break;StaticFunction.sleepIgnoreInterrupt(100);}//returnif (sbReader.length() == 0)return null;elsereturn sbReader.toString();}/*** 关闭所有流*/private void close() {//close outif (out != null) {try {out.close();} catch (IOException e) {e.printStackTrace();}}//errif (err != null) {try {err.close();} catch (IOException e) {e.printStackTrace();}}//inif (in != null) {try {in.close();} catch (IOException e) {e.printStackTrace();}}if (isInReader != null) {try {isInReader.close();} catch (IOException e) {e.printStackTrace();}}if (bInReader != null) {try {bInReader.close();} catch (IOException e) {e.printStackTrace();}}}/*** 销毁*/public void destroy() {//processtry {process.destroy();} catch (Exception ex) {kill(process);}}
}

想了想还是把代码托管到了GitHub上,方便以后分享其他的代码。
地址:Android Utils

非常欢迎大家找出不足发表问题。

[Java][Android][Process] Process 创建+控制+分析 经验浅谈相关推荐

  1. 上海2014科目二注意事项及经验浅谈(龙泉驾校)

    上海2014科目二注意事项及经验浅谈(龙泉驾校) 刚通过科目二考试,其间辛苦与压力,唯有同道之人可知.得益于网络分享,今也总结一番,希望对有需要的人有所帮助. 首先为大家提供倒桩与S弯的方法(这里讲的 ...

  2. SVN使用中的经验浅谈

    上一篇博客简单讲了在合作开发项目时使用SVN的准备工作,而这篇博客则重点在使用中的规范也好,注意事项也好或者使用规则也好.简单说一下使用他的小小经验! 在合作开发项目开始前,贾琳师哥向我们提出了使用S ...

  3. WMS系统条码作业项目实施经验浅谈

    随着市场环境的变化,对现代的企业在物料仓储的仓库,都有了更高的管理要求,以顺应市场的变化,无论是传统的制造企业.贸易公司.电商公司.还是物流供应链公司,都产生了对仓库管理的变革需求,而不再是传统的进销 ...

  4. java中的强制类型转换注意事项_浅谈Java中强制类型转换的问题

    为了更好的理解我们先看下面的例子: package com.yonyou.test; import java.util.ArrayList; import java.util.Iterator; im ...

  5. sybase性能优化经验浅谈

    http://www.2cto.com/database/201204/129094.html 使用sybase数据库多年,有些经验,在数据库部署配置得当,平时监控维护做到位情况下,除此之外数据库的性 ...

  6. 把C#.NET程序移植到DB2上的经验浅谈(C#连接DB2可以用IBM.Data.DB2.dll)

    感谢博客园不再封杀,差点儿搬家的念头都产生了,博客园还是有大胸怀,继续留在博客园写口水文应该没错,娱乐自己.娱乐大家,给枯燥的编程生活增加一点儿笑料,也给大家充当个开心果,让大家高兴一下下.轻松一下下 ...

  7. 数据库设计经验浅谈(3,4,5)转载

    转自:http://bestxiaok.javaeye.com/blog/814870 第 3 部分 - 选择键和索引 数据采掘要预先计划 我所在的某一客户部门一度要处理 8 万多份联系方式,同时填写 ...

  8. 塔菲尔曲线如何分析_浅谈塔菲尔动力学(Tafel Kinetics)

    1.塔菲尔公式 塔菲尔是一个有机化学家,当时他的主要研究集中在通过碳水化合物的还原实现有机物的合成以及有机物的改性,包括己糖,杂环化合物等.在研究过程中,塔菲尔发现一些化合物很难利用传统的同质反应合成 ...

  9. 案例分析:浅谈12306核心模型设计思路和架构设计。

    前言 春节期间,无意中看到一篇文章,文章中讲到12306的业务复杂度远远比淘宝天猫这种电商网站要复杂.后来自己想想,也确实如此.所以,很想挑战一下12306这个系统的核心领域模型的设计.一般的电商网站 ...

  10. java 租户_数据层的多租户浅谈

    在上一篇"浅析多租户在 Java 平台和某些 PaaS 上的实现"中我们谈到了应用层面的多租户架构,涉及到 PaaS.JVM.OS 等,与之相应的是数据层也有多租户的支持. 数据层 ...

最新文章

  1. iOS 排序算法总结、二分法查找
  2. CUDA从入门到精通(零):写在前面
  3. 教你玩转CSS border(边框)
  4. 汇编语言的相对跳转和绝对跳转以及反汇编代码解析
  5. 洛谷-求同构数的个数-NOIP2013提高组复赛
  6. PHP笔记——java程序员看懂PHP程序
  7. 【分享】Android JNI实例​
  8. eplan连接定义点不显示_显示无线桥接成功,但是连接副路由器的设备依然上不去网的问题...
  9. 几个与IP地址类型有关的术语
  10. JAVA程序员必看的15本书-JAVA自学书籍推荐
  11. MonkeyEye电影售票系统--系统顺序图
  12. ☆【平衡二叉树】魔兽争霸
  13. java解析project mpp文件,如何在Java中创建.mpp文件?
  14. JavaPoet 的使用
  15. MATLAB学习笔记(五)——MATLAB绘图
  16. Android SAF(Storage Access Framework)使用攻略
  17. POJ3666(动态规划)
  18. 数码宝贝Butter-fly(罗马音+中文谐音)
  19. MQL4课程-交易函数下单及挂单
  20. 【模型检测学习笔记】10:有限状态迁移系统上的IC3算法

热门文章

  1. 计算机图形学完整笔记(八):曲线曲面 - 2
  2. 如何使用 python 爬取全国小区名称
  3. 主子式大于等于零的矩阵是半正定矩阵的证明方法之二
  4. OllyDBG 入门
  5. 恢复mysql数据--使用frm和ibd文件
  6. CONTINUAL LEARNING FOR AUTOMATED AUDIO CAPTIONING USING THE LEARNING WITHOUT FORGETTING APPROACH
  7. gulp报错task function must be specified
  8. redis_lua_nginx环境配置过程
  9. 51单片机ALE引脚的控制(摘录)
  10. 画图解释 SQL join 语句