Sangsoo Nam | Unlimited Power

Git: Restore a Commit with Reflog

Mon, 25 Jul 2022 00:00:00 +0000

git reset is generally used to cancel a commit. It happens mainly for two reasons. One is just to cancel it. The other is to modify changes. After changes, a new commit can be created. git commit --amend is an alternative way to modify it. Both ways can make the same result.

▶ git log --oneline
a91929a (HEAD -> master) Add skeleton files
6e9be90 Add .gitignore
be44318 Init

▶ git reset HEAD^

▶ git log --oneline
6e9be90 (HEAD -> master) Add .gitignore
be44318 Init

▶ git st
On branch master
Untracked files:
  (use "git add <file>..." to include in what will be committed)
  package.json
  yarn.lock
  app.tsx
  index.html

nothing added to commit but untracked files present (use "git add" to track)

These commands are quite useful and powerful. The thing is that I sometimes want to cancel those and restore the previous commit. After git reset, it is gone. git log doesn’t show it anymore.

▶ git log --oneline
6e9be90 (HEAD -> master) Add .gitignore
be44318 Init

Then, what would be the way to restore it? Maybe you pushed a branch or a commit to a remote repository before. When git reset happens only in a local machine, it’s possible to get or restore it from the remote repository. If not, it can be tricky.. Luckily, git has reference logs. The git document describes

Reference logs, or “reflogs”, record when the tips of branches and other references were updated in the local repository.

Basically, git doesn’t remove commits right away when git reset happen. It changes references first. In this example, HEAD is now changed to 6e9be90. git reflog shows those reference change logs. These logs are stored only locally and git will prune reference logs older than 90 days by default. The below command shows the commit id, a91929a, before running git reset HEAD^.

▶ git reflog
6e9be90 (HEAD -> master) HEAD@{0}: reset: moving to HEAD^
a91929a HEAD@{1}: commit: Add skeleton files
6e9be90 (HEAD -> master) HEAD@{2}: commit: Add .gitignore
be44318 HEAD@{3}: commit (initial): Init

git reset --hard <COMMIT_ID> would restore that commit and references.

▶ git reset --hard a91929a
HEAD is now at a91929a Add skeleton files

▶ git log --oneline
a91929a (HEAD -> master) Add skeleton files
6e9be90 Add .gitignore
be44318 Init

The same way is applicable to restore after git commit --amend. This command is used to apply more changes to the existing commit. Technically, git commit --amend does’t modify an existing commit. It will create a new commit using the changes of the most recent existing commit. Then, git adjust references to show only the new commit. git log doesn’t show that previously existing commit.

▶ git commit --amend
[master 447608a] Add skeleton files
 ...

▶ git log --oneline
447608a (HEAD -> master) Add skeleton files
6e9be90 Add .gitignore
be44318 Init

It means that it would be easy to restore by knowing that previously existing commit id. git reflog shows those records. The commit id before amending is a91929a.

▶ git reflog
447608a (HEAD -> master) HEAD@{0}: commit (amend): Add skeleton files
a91929a HEAD@{1}: reset: moving to a91929a
6e9be90 HEAD@{2}: reset: moving to HEAD^
a91929a HEAD@{3}: commit: Add skeleton files
6e9be90 HEAD@{4}: commit: Add .gitignore
be44318 HEAD@{5}: commit (initial): Init

git reset --hard command will restore that commit and references.

▶ git reset --hard a91929a
HEAD is now at a91929a Add skeleton files

▶ git log --oneline
a91929a (HEAD -> master) Add skeleton files
6e9be90 Add .gitignore
be44318 Init

One more thing. If you want to learn more on references, try git cat-file -p <COMMIT_ID|BRANCH>.

# After `git reset --hard a91929a`, the references are the same as `a91929a`
▶ git cat-file -p master
tree 3f5d073729ffc62a7a1f9e307ba510a3f4762ea9
parent 6e9be90d94f41e176f1a5a9a706e14fdb10f1586

# `a91929a` have a parent, `6e9be90`
▶ git cat-file -p a91929a
tree 3f5d073729ffc62a7a1f9e307ba510a3f4762ea9
parent 6e9be90d94f41e176f1a5a9a706e14fdb10f1586

# Previously amend commit has the same parent, `6e89e90`
▶ git cat-file -p 447608a
tree 3f5d073729ffc62a7a1f9e307ba510a3f4762ea9
parent 6e9be90d94f41e176f1a5a9a706e14fdb10f1586

Bash: Highlight Matched Texts With Colors

Thu, 24 Dec 2020 00:00:00 +0000

Logs are valuable sources to identify issues. It would be continuously generated (e.g. adb logcat) or stored as a file. Mostly, I use the terminal to check those. When I check that, it is often not easy to see since every text has the same color. grep could help to filter out. With --color=auto or --color=always option, it supports highlighting texts as well. However, there are cases I don’t want to filter out but want to highlight certain texts or words to check easily. This happens more frequently when I check continuous logs.

After some investigation, I’ve figured out how to do it in bash. By adding the below function, you can highlight a matched text with a specified color.

.bash_profile

highlight() {
  # Usage: highlight <color> <text>
  #
  # Foreground colors
  # ---
  # 30:black, 31:red, 32:green, 33:yellow, 34:blue, 35:magenta, 36:cyan

  # Background colors
  # ---
  # 40:black, 41:red, 42:green, 43:yellow, 44:blue, 45:magenta, 46:cyan
  escape=$(printf '\033')
  sed "s,$2,${escape}[$1m&${escape}[0m,g"
}

The actual showing color is determined by terminal settings. For example, iTerm defines colors in Profiles > Colors.

Here is an example on how to use highlight and how the result looks.

$ echo Hello | highlight 31 H | highlight 32 e | highlight 33 l | highlight 34 o

Though this function works well, it requires quite a long command if I want to highlight multiple texts, as you can see above. Specifying a color is great, but it becomes somewhat cumbersome if it is needed every time. All I need is just to highlight texts with different colors. Based on the highlight function, the below helper function simplifies it.

.bash_profile

ht() {
  if [[ $# == 1 ]]; then
    highlight 31 $1
  fi
  if [[ $# == 2 ]]; then
    highlight 31 $1 | highlight 32 $2
  fi
  if [[ $# == 3 ]]; then
    highlight 31 $1 | highlight 32 $2 | highlight 33 $3
  fi
  if [[ $# == 4 ]]; then
    highlight 31 $1 | highlight 32 $2 | highlight 33 $3 | highlight 34 $4
  fi
}

Failing Unit Test by GeneratedSerializationConstructorAccessor1

Thu, 15 Aug 2019 00:00:00 +0000

For running Gradle tasks, I prefer using a command-line interface. Recently, I made a code change and added new unit tests. To verify the behavior, I ran a test task. Soon after, it showed a quite weird error message, which was about GeneratedSerializationConstructorAccessor1. As you can see below, the error message didn’t point out any line from my code. Everything was related to jdk. When I made a pull request, TeamCity ran the test task without any issue. It made me curious and I wanted to solve this issue.

$ ./gradlew test
> Task :app:testDebugUnitTest
...
java.lang.NoClassDefFoundError: jdk/internal/reflect/GeneratedSerializationConstructorAccessor1
        at jdk.internal.reflect.GeneratedSerializationConstructorAccessor1.newInstance(Unknown Source)
        at java.base/java.lang.reflect.Constructor.newInstanceWithCaller(Constructor.java:500)
        at java.base/java.lang.reflect.Constructor.newInstance(Constructor.java:481)
        at java.base/java.io.ObjectStreamClass.newInstance(ObjectStreamClass.java:1081)
        at java.base/java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:2071)
        at java.base/java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1594)
        at java.base/java.io.ObjectInputStream.readObject(ObjectInputStream.java:430)
        at org.gradle.process.internal.worker.child.SystemApplicationClassLoaderWorker.call(SystemApplicationClassLoaderWorker.java:107)
        at org.gradle.process.internal.worker.child.SystemApplicationClassLoaderWorker.call(SystemApplicationClassLoaderWorker.java:64)
        at worker.org.gradle.process.internal.worker.GradleWorkerMain.run(GradleWorkerMain.java:62)
        at worker.org.gradle.process.internal.worker.GradleWorkerMain.main(GradleWorkerMain.java:67)
Caused by: java.lang.ClassNotFoundException: jdk.internal.reflect.GeneratedSerializationConstructorAccessor1
        at java.base/java.net.URLClassLoader.findClass(URLClassLoader.java:436)
        at java.base/java.lang.ClassLoader.loadClass(ClassLoader.java:588)
        at java.base/java.lang.ClassLoader.loadClass(ClassLoader.java:521)
        ... 11 more
...

The error message could vary slightly depending on your Jacoco tool version.

This error message showed some lines in jdk. Since there was no issue in TeamCity, it might be a bug in jdk and possibly I used a different jdk version from what TeamCity used.

$ java -version
java version "12.0.2" 2019-07-16
Java(TM) SE Runtime Environment (build 12.0.2+10)
Java HotSpot(TM) 64-Bit Server VM (build 12.0.2+10, mixed mode, sharing)

TeamCity used Java 8 and my machine used Java 12. I updated my Java version to work on backend services, which now requires higher that Java 11. After updating, Gradle seemed work well. I could run the assemble task. The problem only happened for the test task strangely.

After researching deeply, this turned out a quite complex problem. In short, Java 8 and higher versions (e.g. Java 9, Java 10, …) are different. Plugins or libraries should support higher versions to work properly. For me, they are Jacoco, Mockito, and PowerMock. There are quite typical plugins/libraries for Android testing.

For Jacoco, you can fix this issue by updating the plugin version and adding jacoco.excludes.(Gradle issue #5184).

jacoco {
    toolVersion = '0.8.4'
}

tasks.withType(Test) {
    jacoco.includeNoLocationClasses = true
    jacoco.excludes = ['jdk.internal.*']
}

Mockito [had an update (Mockito PR #1250) to support this. If you see below error, you need to update mockito-core to use the latest version. The current project used 2.8.9 version which is quite old and didn’t have that update. The latest Mockito version at this point is 2.28.2 and 3.0.0.

$ ./gradlew test
> Task :app:testDebugUnitTest
...
com.project.ExampleTest > basicFunctionality FAILED
    org.mockito.exceptions.base.MockitoException at ExampleTest.java:36
        Caused by: java.lang.UnsupportedOperationException at ExampleTest.java:36
            Caused by: java.lang.IllegalArgumentException at ExampleTest.java:36
...

For me, there was another hurdle. It still didn’t work and it was due to PowerMock. Unfortunately, it seemed not fixed yet (Powermock issue #901).

$ ./gradlew test
> Task :app:testDebugUnitTest
...
com.project.ExampleText > initializationError FAILED
    org.objenesis.ObjenesisException
        Caused by: java.lang.reflect.InvocationTargetException
            Caused by: java.lang.IllegalAccessError

...

If you don’t use PowerMock library, updating libraries and setting excludes of Jacoco would fix your issue. In my case, however, it was tricky because of PowerMock. There might be some way to work around but it seemed not worth spending more time. As you know, the issue was caused by jdk. It means that this could be easily solved by using Java 8 instead of Java 12. jenv(https://www.jenv.be) helps to manage/switch multiple jdks in the same machine. Using that, I was able to fix all my issue and to switch Java 12 when I need to work on the backend code.

$ jenv add /Library/Java/JavaVirtualMachines/jdk1.8.0_221.jdk/Contents/Home
$ jenv versions
  oracle64-1.8.0.221
* oracle64-12.0.2 (set by /Users/sangsoo/.jenv/version)
$ jenv global oracle64-1.8.0.221

If you use Android Studio to run gradle tasks including test, you might not see this issue. Android Studio allows you to set jdk. It could be different from your command line environment. By default, Android Studio uses an embedded JDK, which is Java 8.

Why ZSH is Slow for Git Repos

Mon, 05 Aug 2019 00:00:00 +0000

In these days, I’m using ZSH(https://ohmyz.sh/). It has many handy features and supports useful plugins. One day, I’ve noticed a slowness issue. Basically, the prompt was not shown right after running a command. It was interesting because it only happened in the git repositories. To fix this issue, I needed to understand what’s happening first.

When you run zsh with a --help parameter, you can see many options. Among them, -v and -x are useful ones to dig into the problem.

$ zsh --help
...
-v    equivalent to --verbose
-x    equivalent to --xtrace
...

When you run zsh -xv, every command that ZSH runs will be shown. In the git repo, I pressed enter multiple times and checked logs to figure out which command is slow. There might be some better ways but I just looked at the terminal. I saw that there was a bit freeze when the below line was shown.

$ zsh -vx
...
+git_prompt_status:1> local INDEX STATUS
+git_prompt_status:2> INDEX=+git_prompt_status:2> git status --porcelain -b
...

To show helpful information on the prompt, ZSH has many defined functions. Git related functions are defined in git.zsh. As you can see, that command is from git_prompt_status().

.oh-my-zsh/lib/git.zsh

...
# Get the status of the working tree
function git_prompt_status() {
  local INDEX STATUS
  INDEX=$(command git status --porcelain -b 2> /dev/null)
  STATUS=""
...

Those functions are generally called by ZSH themes and plugins. In my case, avit theme, I’m currently using, uses it to set RPROMPT.

.oh-my-zsh/themes/avit.zsh-theme

...
RPROMPT='$(_vi_status){$(echotc UP 1)%}$(_git_time_since_commit) $(git_prompt_status) ${_return_status}%{$(echotc DO 1)%}'
...

It is possible to fix or optimize git_prompt_status() function in git.zsh. However, I realized that it was the information that I didn’t need actually. So, I decided to remove that. One way is to remove that line in the theme directory. .oh_my_zsh is a git repo. So, modifying the code in there could make an issue to update ZSH. Instead of that, I simply override RPROMPT value in .zshrc. This is a safer approach.

.zshrc

...
RPROMPT=''
...

Note that a problem line could vary depending on your theme and plugins. When you have some issues, zsh -vx will help you to point a problem command.

How to Embed a Google Drive Video on a Website

Sat, 27 Jul 2019 00:00:00 +0000

There are lots of video sites you can upload your videos. After uploading, you can easily embed that video on your post or on a website. For example, YouTube supports “Copy embed code” option when you do the right-click.

The copied code is like below. iframe tag is used to embed a video, and there are other parameters such as allowfullscreen.

<iframe width="766" height="431" src="https://www.youtube.com/embed/hVGVcmW0crs"
frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope;
picture-in-picture" allowfullscreen></iframe>

As you can see, it is easy to embed videos on your site. Although there are well-known services, you might consider embedding a Google drive video rather than using them. There are several benefits. When you upload your video, most sites consider that as public. It means that you might not be able to specify who you want to show. When you share something from Google drive, you can easily set who can see. It also has a handy permission scope, an organization.

That being said, embedding a Google drive video supports a better permission scope and security. If you consider it, here are the steps. After uploading your video, you first need to share it. You can set specific users or an organization.

Unlike YouTube, finding “Embed code” in Google drive is not straightforward. First of all, you need to click “Preview”.

Then, open the menu by pressing the top-right icon, and click “Open in new window”. That will open a new tab.

When you open the menu again in that tab, you will see the option, “Embed item…“.

Finally, you can copy the embed code and paste it on your site.

Note that full screen mode is not allowed if you just copy and paste that code. You can allow it by adding allowfullscreen in the iframe tag.

<iframe src="https://drive.google.com/file/d/1rLT6EI-example/preview"
width="640" height="480" allowfullscreen></iframe>

Adb: Take a Screenshot with a Laptop

Fri, 19 Jul 2019 00:00:00 +0000

When you use an Android emulator, taking a screenshot is quite easy. Just press the ‘Take screenshot’ on the sidebar or press the shortcut (For Mac, ⌘ + S). However, it is slightly cumbersome if you want to take a screenshot from your device and store that in your laptop. The reason is that you need to pull a screenshot file after taking a screenshot in the device.

When the device is connected to your laptop, you can do it with adb commands: shell and pull. The first command takes a screenshot and store into your device. The second command pulls it to your laptop from the Android device. Although it works, it makes a bit unnecessary intermediate file in your device, which is screenshot.png.

$ adb shell screencap -p /sdcard/screenshot.png
$ adb pull /sdcard/screenshot.png

In addition to an intermediate file, this requires two commands. This is not a big deal but you may want to find a simpler way. If you are looking for it, below command is it. This command takes a screenshot from the connected device and stores it on your laptop.

$ adb exec-out screencap -p > screenshot.png

This command saves a screenshot to screenshot.png. As you notice, this will overwrite an existing file when you run multiple times. You probably don’t want to do. When you click ‘Take screenshot’ on Android emulator, it uses a naming convention, Screenshot_{unix_timestamp}.png. To do the same behavior, you can utilize date command.

$ adb exec-out screencap -p > `date +Screenshot_%s.png`

If you run this frequently, you can set an alias for that.

.zshrc

alias asc='adb exec-out screencap -p > `date +Screenshot_%s.png`'

This command is enough for most cases. However, there is still some chance to overwrite because unix_timestamp uses seconds. If you concern that, you can add a millisecond additionally. %N returns nanoseconds and %3N returns milliseconds by cutting the first 3 digits.

$ adb exec-out screencap -p > `date +Screenshot_%s.%3N.png`

Unfortunately, date command in the mac doesn’t support %N. So, you can’t get milliseconds. If you have this issue, use gdate(GNU date) command.

$ brew install coreutils
$ adb exec-out screencap -p > `gdate +Screenshot_%s.%3N.png`

brew install coreutils installs gdate on Mac.

RxJava: Fail Early and Track Errors

Sun, 14 Jul 2019 00:00:00 +0000

In RxJava, Publisher emits events and you can subscribe to those events with Consumer. While subscribing, emitted events can be used to update the view or to do something else. Below example shows a simple example:

getUser() // Publisher. Events could be from the network or database.
  .subscribeOn(Schedulers.io())
  .observeOn(AndroidSchedulers.mainThread())
  .subscribe(user ->  {
      // Consumer
      userTextView.setText(user.getName());
    }
  });

This code work well while Publisher emits events without any error. However, there could be several issues to emit events. Those vary depending on where events are generated. For example, if Publisher is related to the network, it could have a bad connection or no response from a backend service. In that situation, an error will be thrown. The current code doesn’t have any logic to consume that error, OnErrorMissingConsumer exception happens. That exception makes an app crash.

The crash will tell you what went wrong. Although it is useful, you don’t want to make a crash in some cases. Some logic doesn’t affect the main experience, or a user can trigger to subscribe again easily. So, you can set an error consumer to handle those errors and avoid a crash. One way is just to print an error log.

getUser() // Publisher. Events could be from the network or database.
  .subscribeOn(Schedulers.io())
  .observeOn(AndroidSchedulers.mainThread())
  .subscribe(user ->  {
      // Event Consumer
      userTextView.setText(user.getName());
    }, throwable -> {
      // Error Consumer
      Log.e(TAG, "Failed to get the user name");
    }
  });

It looks good, but it has some potential issues.

This doesn’t make a crash, so it is easy to miss. Error logs are often not enough to grab attention.
There is no way to know how many people have errors. If your backend service is down, all users will have a network error. By tracking errors, you can keep giving a better experience continuously.

Thus, what we want to achieve is to track errors in the release and to make a crash while developing, which helps early detection. Crashlytics is one of the useful tools for tracking errors. You can track errors by calling Crashlytics.logException.

Crashlytics.logException(throwable);

BuildConfig.DEBUG tells whether the build is a release or a debug. TO sum up, you can resolve those issues like below.

...
    }, throwable -> {
      // Error Consumer
      RxJavaErrorHandler.handle(throwable);
    }
...

public class RxJavaErrorHandler {

    public static void handle(Throwable throwable) {
        if (BuildConfig.DEBUG) {
            Thread.currentThread()
                    .getUncaughtExceptionHandler()
                    .uncaughtException(Thread.currentThread(), throwable);
        } else {s

            Crashlytics.logException(throwable);
        }
    }
}

Some exceptions/throwables could be not needed to make a crash in your application. You can check with instanceof and ignore them.

Use the <merge> Tag for Optimizing View Tree

Wed, 20 Mar 2019 00:00:00 +0000

In Android, a view is one of the basic blocks for user interface components. It is a tree structure and you can nest views to display what you want to show. If the tree height gets longer, it can lead to some performance issue. The reason is that the Android framework does a top-down traversal of the view tree two times: a measure pass and a layout pass. Thus, it is important to keep in mind to make the view tree height short as much as possible.

FrameLayout is one of the views, and it allows placements of views along the z-axis. So, you can stack view elements. In addition that, you can align views with a layout_gravity attribute. Here is an example.

<?xml version="1.0" encoding="utf-8"?>
<FrameLayout xmlns:android="http://schemas.android.com/apk/res/android"
    android:layout_width="match_parent"
    android:layout_height="match_parent">

    <TextView
        android:layout_width="wrap_content"
        android:layout_height="wrap_content"
        android:layout_gravity="center"
        android:textSize="20sp"
        android:text="Hello World!" />

    <Button
        android:layout_width="wrap_content"
        android:layout_height="wrap_content"
        android:layout_gravity="bottom|center"
        android:text="Button" />

</FrameLayout>

While running the application, you can capture the view tree. Click Tools > Layout Inspector. Then, you can see the below screen.

In the layout file, FrameLayout is the root view. That is true if you see that layout file. When you look closely at the view tree, you can find ContentFrameLayout, the parent of your FrameLayout. Android framework creates it internally and adds your views in it. It is a FrameLayout with a different name.

public class ContentFrameLayout extends FrameLayout {
  public ContentFrameLayout(Context context) {
      this(context, null);
  }

  public ContentFrameLayout(Context context, AttributeSet attrs) {
      this(context, attrs, 0);
  }

  public ContentFrameLayout(Context context, AttributeSet attrs, int defStyleAttr) {
      super(context, attrs, defStyleAttr);
  }
}

If that is almost the same, FrameLayout in your layout file seems unnecessary. As you know, it is better to have a short view tree for the performance. You can do it with the <merge> tag. The <merge> tag is designed to eliminate redundant view groups. Android framework places views in the <merge> tag directly into the parent layout. The layout file can be changed like this.

Since the parent layout is used, you can remove layout_width and layout_height attributes. Actually, any attributes in <merge> tag are just ignored even though you set.

<?xml version="1.0" encoding="utf-8"?>
<merge xmlns:android="http://schemas.android.com/apk/res/android">

    <TextView
        android:layout_width="wrap_content"
        android:layout_height="wrap_content"
        android:layout_gravity="center"
        android:textSize="20sp"
        android:text="Hello World!" />

    <Button
        android:layout_width="wrap_content"
        android:layout_height="wrap_content"
        android:layout_gravity="bottom|center"
        android:text="Button" />

</merge>

Conclusion

A shorter view tree brings better performance in Android. Android frame supports <merge> to eliminate redundancy view groups. It is useful especially when you use FrameLayout as a root view. With that, you can use the ContentFrameLayout directly.

How to Set Default Values in Room Database

Fri, 15 Mar 2019 00:00:00 +0000

When you initialize your application, you often need a default value to determine how it works. For example, most values of Settings require a default value. When you use a database, you read a value from there. Thus, it is better to put a default value in there instead of in the code.

How can you do it? Room database supports a callback method. There are two callbacks: onCreate and onOpen. onCreate is called only when the database is created for the first time. After that, this is not called again. onOpen is called when the database is opened. Both callbacks are not called when your application is started but called when your code accesses the database such as querying.

public abstract static class Callback {

  public void onCreate(@NonNull SupportSQLiteDatabase db) {
  }

  public void onOpen(@NonNull SupportSQLiteDatabase db) {
  }
}

If you want to pre-populate data only one time, you can do it using onCreate callback. However, it wouldn’t fit well if there is a chance to add another default value later. If you think about Settings, there could be a new setting option in the next version.

When the application is restarted, onOpen can be called unlike onCreate. Now the remaining question is how to set default values properly. Since onOpen can be called several times, you need to pre-populate data only if it is necessary. If not, you could end up overriding data changed by a user.

This sounds like a bit tricky but actually quite simple to do. Room database supports to select a conflict algorithm. By selecting, CONFLICT_IGNORE, you can set default values without worrying overriding. Room database doesn’t allow to access the database on the main thread unless you set allowMainThreadQueries(). Thus, the callback method is executed not on the main thread.

final AppDatabase database = Room.databaseBuilder(this, AppDatabase.class, "database.db")
  .addCallback(new RoomDatabase.Callback() {

    @Override
    public void onOpen(@NonNull SupportSQLiteDatabase db) {
      super.onOpen(db);
      Map<String, String> defaultMaps = new HashMap<>();
      defaultMaps.put("theme", "dark");
      defaultMaps.put("auto_play_mode", "true");
      defaultMaps.put("shuffle_mode", "true");

      for (Map.Entry<String, String> entry : defaultMaps.entrySet()) {
        ContentValues values = new ContentValues();
        values.put("key", entry.getKey());
        values.put("value", entry.getValue());
        db.insert("Setting", SQLiteDatabase.CONFLICT_IGNORE, values);
      }
    }
  })
  .build();

Android: Analyze Stack Trace from Plain Text Log

Sun, 10 Mar 2019 00:00:00 +0000

A stack track is a report that shows the stack frames at a certain point. When you debug your application, this is useful to know how to this method is called. This is even useful when a crash happens. At that time, you need to know what happened. A stack track can show how to end up having a crash. By analyzing it, you can find the root problem in your application.

When you have a crash issue while developing, you can see it in the Logcat tab in Android Studio. In the log, there are blue and grey texts, which tell you the file and line information on that stack frame. The blue means that stack frame is in your source code. Whereas, the grey means that it is in a third-party library or framework code. By clicking it, you can easily navigate.

Crashes can happen not only when you are using Android Studio but also when anytime. Someone can report an issue with a log. Or, you can see them on the crash reporting tool such as Fabric. For both cases, you will see the plain text log like below since that didn’t happen in the Android Studio.

java.lang.NullPointerException: Attempt to invoke virtual method 'boolean com.todoapp.data.Task.isEmpty()' on a null object reference
    at com.todoapp.addedittask.AddEditTaskPresenter.createTask(AddEditTaskPresenter.java:142)
    at com.todoapp.addedittask.AddEditTaskPresenter.saveTask(AddEditTaskPresenter.java:90)
    at com.todoapp.addedittask.AddEditTaskFragment.lambda$onActivityCreated$0(AddEditTaskFragment.java:75)

From this log, you can navigate the file and line manually and one by one. If you want to see a below stack frame, you can open another file and go to that line. This is absolutely doable but a bit cumbersome. When a crash happens while developing, it is really easy to navigate since you just need to click. For that plain text log, Android Studio supports to convert it clickable after parsing. Open Analyze > Analyze Stack Trace... menu. Then, paste the log and click OK.

The parsed stack trace will be shown and you can now easily navigate. If you analyze the plain text log a lot, you can turn on the Automatically detect and analyze option. With that, Android Studio show the parsed stack trace without opening the menu.