Gokuldroid RSS Feed

Find disk usage sorted by size

Tue, 11 Jul 2023 17:26:02 GMT

To find all folders and it's disk usage sorted by size, you can use the following cmd

Copy

du -ck | sort -n

To find all files in a folder and it's size,

Copy

ls -l --block-size=MB

Copy

ls -lh

Find process using a port

Thu, 06 Jul 2023 23:27:48 GMT

you can execute the following cmd to find the process which occupies a specific port in linux.

Copy

sudo lsof -i :<port_number> -t | xargs ps -o pid,user,cmd -p

Top 10 tools you need as developer

Sat, 11 Mar 2023 13:07:07 GMT

Introduction

As a developer, it is essential to have the right set of tools to help you streamline your workflow and boost productivity. With the plethora of options available, it is easy to get overwhelmed. In this blog post, we will discuss the top 10 tools every developer needs to have to make their lives easier.

1 - A Great editor / IDE

Most of the time you will be spending your time with your code editor. thus you have to choose your main weapon and you should know how to use it effectively.everyone will have their personal preferences when it comes to code editor. you should try few and settle with what you're comfortable. I would also suggest you to learn and commandline based editor like VIM. that will be handy when you're dealing with ssh / remote machines etc. Atleast you should know how to exit VIM :P.

personal recommendations:

VS code (It has great extensibility. If you know typescript I would recommend this, since you can write your own plugins for your use case).
Sublime text (Blazing fast one).
Intellij (Swiss army knife).
VIM (Hard kid).

2 - Version control system - Git

Git is one of the most powertool that you can learn as a developer. you can learn about git and it's internals here. Some people will be overwhelmed with the commands that git can provide. You can use any git client like VS code (git lens plugin). I would also recommend you to setup autocompletions for git if you prefer to use cmdline.

3 - A good note taking app

A good note taking app will save you tone of time. I personally use notion / bear notes. Notion takes time to learn. but once you get hang of it, you will be unstoppable. I use bear notes for it's simplicity. It uses plan old markdown. thus it makes it easier to take notes. both comes with syntax highlighting, that's a must for people like us.

other recommendations:

Evernote
Microsoft's Onenote
Apple notes
Google keep

4 - A good terminal

Teh default terminal comes with mac and windows is not that great. you might want to consider cmder incase if you're using windows. iTerm2 + zsh if you're from MAC world.

5 - Raycast - good all in one command prompt (Mac only).

Raycast is much powerful than the the spotlight search in mac. It has tons of plugins to get started with. It has floating notes, clipboard history etc. you will never regret tying this.

6 - Commandline json processor

If you're dealing with lot of huge json content in your work, JQ will be your best friend. you can extract / search / modify json with ease.

7 - Incase if you hate cmdline and you have no choice

navi can eliminate your pain of knowing lots of cmds. creating your own scripts takes time but if that clubbed with fzf it can do amazing things. It also comes with lot of plugins to deal with common tools like git, aws, azure database etc.

8 - Postman - API testing

Postman is a powerful API testing and development tool that makes it easy to create, test, and document APIs. With Postman, you can send requests and receive responses from APIs, create automated tests, and generate documentation for your APIs. It also provides a user-friendly interface for managing and organizing your APIs, making it a must-have tool for developers working with APIs.

9 - Docker

Now a days docker is standard tool setting up dev environment. Your application might be using different different version of sdks. each might have it's own environment etc. handling these via docker makes it easy for everyone in your team to have a setup which will work out of the box.

10 - Database viewer

Most of us will deal with databases in our day to day life. having a good database viewer will save you tones of time. dbeaver is an all in one database tool that can you help you write / execute queries and view results. it supports wide range of databases like MySQL, PostgreSQL, SQLite, Oracle, DB2, SQL Server, Sybase, MS Access, Teradata, Firebird, Apache Hive, Phoenix, Presto, etc.

mhowls (Must have or we look stupid)

Fri, 27 Jan 2023 10:34:41 GMT

"Must have or we look stupid" is applicable all the tasks that you do, all the questions that you ask to everyone. Say you're developing a login / signup page. at base minimum it should handle positive login and negative login cases. extensive validations, unique username check comes after those first two cases. Say all the testing is completed and we launched it to production. As a first step QA engineer will do a sanity of those bare minimum cases. if those didn't work, ultimately we will be stupid in-front of your customers right?. but have we thought about this while we ask questions to others. before asking questions to our seniors or fellow developers, we should have explored / tried very obvious solutions first ourselves. most of your team mates will be helpful even if you ask stupid questions. but doing basic exploration is the justice we can do those helpful people. and also yaa, we don't need to look stupid after asking a silly question. be a QA engineer for your own questions.

Being a privileged website

Mon, 17 Oct 2022 09:03:29 GMT

We all know that not everyone in this world is having equal privileges. I have spent quite a lot of time fixing IE bugs in my early stage of my career. It is super frustrating to handle all edge cases for all supported versions of every famous browsers. But do you believe that even websites doesn't have equal privileges in the browsers that we use in our day to day life?. Yes, it is true. the same code and same business logic might not work in the exact same browser, if you deploy it in the different domain. read more to understand what I am talking about

you might know webkit already if you're frontend developer. incase if you don't know about it already.

WebKit is a cross-platform web browser engine. On iOS and macOS, it powers Safari, Mail, iBooks, and many other applications.

Recently, apple made it open source in github. by digging at it's source we can find some interesting stuffs.

For example this one, on a hindsight it tries to suppress native auto correction in google docs. but why does even a browser engine developed by apple originally having some hacks to support a feature developed by google. surprising isn't it?.

Copy

static bool shouldSuppressAutocorrectionAndAutocaptializationInHiddenEditableAreasForHost(StringView host)
{
#if PLATFORM(IOS_FAMILY)
    return equalLettersIgnoringASCIICase(host, "docs.google.com"_s);
#else
    UNUSED_PARAM(host);
    return false;
#endif
}

Need more?

here are some examples,

Copy

// FIXME: Remove after the site is fixed, <rdar://problem/75792913>
bool Quirks::shouldHideSearchFieldResultsButton() const
{
#if ENABLE(IOS_FORM_CONTROL_REFRESH)
    if (!needsQuirks())
        return false;

    if (topPrivatelyControlledDomain(m_document->topDocument().url().host().toString()).startsWith("google."_s))
        return true;
#endif
    return false;
}

They even have special handling with CSS class name.

Copy

bool Quirks::shouldMakeTouchEventNonCancelableForTarget(EventTarget* target) const
{
    if (!needsQuirks())
        return false;

    auto host = m_document->topDocument().url().host();

    if (equalLettersIgnoringASCIICase(host, "www.youtube.com"_s)) {
        if (is<Element>(target)) {
            unsigned depth = 3;
            for (auto* element = downcast<Element>(target); element && depth; element = element->parentElement(), --depth) {
                if (element->localName() == "paper-item"_s && element->classList().contains("yt-dropdown-menu"_s))
                    return true;
            }
        }
    }

    return false;
}

And this one,

Copy

static bool isKinjaLoginAvatarElement(const Element& element)
{
    // The click event handler has been found to trigger on a div or
    // span with these class names, or the svg, or the svg's path.
    if (element.hasClass()) {
        auto& classNames = element.classNames();
        if (classNames.contains("js_switch-to-burner-login"_s)
            || classNames.contains("js_header-userbutton"_s)
            || classNames.contains("sc-1il3uru-3"_s) || classNames.contains("cIhKfd"_s)
            || classNames.contains("iyvn34-0"_s) || classNames.contains("bYIjtl"_s))
            return true;
    }

    const Element* svgElement = nullptr;
    if (is<SVGSVGElement>(element))
        svgElement = &element;
    else if (is<SVGPathElement>(element) && is<SVGSVGElement>(element.parentElement()))
        svgElement = element.parentElement();

    if (svgElement && svgElement->hasAttributes()) {
        auto ariaLabelAttr = svgElement->attributes().getNamedItem("aria-label"_s);
        if (ariaLabelAttr && ariaLabelAttr->value() == "UserFilled icon"_s)
            return true;
    }

    return false;
}

A sample PR description of a frustrated developer

Copy

 Regression(252759@main) Unable to log into pandora.com

https://bugs.webkit.org/show_bug.cgi?id=246430
rdar://100243111

Reviewed by Geoffrey Garen and Ryosuke Niwa.

We're unable to log into pandora.com since un-exposing window.showModalDialog()
in 252759@main. Pandora does not actually seem to call showModalDialog() but
they are calling its getter on the Window object. It is unclear why this is
causing log in to fail since there is no JS error (and other browsers don't
expose showModalDialog either).

To resolve the issue for now, I am adding a quirk to re-expose showModalDialog
on pandora.com. However, since they're not calling it, I am merely using
`undefined` as value for the property.

you can find full source code for these hacks here. If you ever find a bug in your site that doesn't work for some reason, don't try fix it in your code. fix it in the browser engine itself like a pro frontend developer and have a life with your pet dog or GF. Happy monday frontend folks :).

Ostrich algorithm

Sun, 02 Oct 2022 07:58:25 GMT

With ever growing tech debts in any software products, sometimes we are overwhelmed with the number of backlog items. let's say we have a email validation in website during sign-up. we might want to do uniqueness and email format and max length of email according the RFC standard. we can only prevent case one by invoking backend. but we can prevent user inputting case two and three with plain javascript code. even if we forgot to add length validation, we will validate the email ownership by sending a mail to the inputted email. at worst case we will send invite link to invalid mail. according to wikipedia,

the ostrich algorithm is a strategy of ignoring potential problems on the basis that they may be exceedingly rare. It is named after the ostrich effect which is defined as "to stick one's head in the sand and pretend there is no problem". It is used when it is more cost-effective to allow the problem to occur than to attempt its prevention.

we can very well ignore the length validation since, anyways it will be validated in some other means. although email address RFC standard most likely won't change, maintaining the RFC standard based validation in our code is still and overhead. we can be ostrich in this case and pretend there is no problem.

Insert current time in vs code using snippet

Thu, 15 Sep 2022 22:16:11 GMT

Cmd + shift + p -> configure user snippets -> new global snippets -> paste the following snippet -> save -> and type currenttime

Copy

{
	"current time": {
		"prefix": "currenttime",
		"body": [
			"$CURRENT_YEAR-$CURRENT_MONTH-${CURRENT_DATE}T$CURRENT_HOUR:$CURRENT_MINUTE:$CURRENT_SECOND.000Z"
		],
		"description": "Insert current time"
	}
}

Supported snippet variables can be found here. More about vs code snippets.

Setting up github custom domain

Sun, 11 Sep 2022 12:30:34 GMT

Setting up a custom domain for our github pages has lot of benefits but at the same time it requires lot of patience. This article aims to give a picture about how the entire thing works and step-by-step guide for the same. I am taking my github profile Gokuldroid and my custom domain codefromdude.com as example here.

Reasons on why you need to setup custom domain for your github page

The default domain provided by github <username.github.io> is not intuitive.

We will lose all google indexing if we want to move away from github.

For more personal branding.

And we're devs and we want problems ;-).

Prerequisites

Github account

Cloudflare account (github doesn't support SSL termination for custom domains. we can use cloudflare to do the SSL termination for our blog for free)

A repository (.github.io) with two files (index.html and CNAME)

A domain registered in any providers (Ex, Godaddy, Namecheap, Cloudflare etc)

Bit of patience.

Setting up your github repository

(skip this if you already have site published at <github-user-name>.github.io)

First you have to create a public repository in this format (<github-user-name>.github.io -> gokuldroid.github.io)
clone github repo locally using this command and cd into the folder and create index.html file

Copy

git clone https://github.com/<your-github-username>/<your-github-username>.github.io
cd <your-github-username>.github.io
touch index.html

Add a index.html in the root folder

Copy

<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta http-equiv="X-UA-Compatible" content="IE=edge">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>Document</title>
</head>
<body>
    <p>Hello world</p>
</body>
</html>

Push the changes to remote

Copy

git add --all
git commit -m "Initial commit"
git push -u origin main

Fire up a browser and go to https://<your-github-username>.github.io. Now you have successfully setup a personal page in github.

Setting up domain in cloudflare and github

create CNAME file with your custom domain (should be in single line without any spaces. reference).

Copy

cd <your-repository>
echo "<your-custom-domain>" > CNAME
git add --all
git commit -m "Add CNAME file"
git push -u origin main

Go to github profile settings -> Code, planning, and automation -> Pages -> click on Add domain
Type your custom domain name -> click Add domain. This will take you to the following the page like this. we need this txt record and value for the next step. keep this handy.

Go to cloudflare -> websites -> Add website -> Type your custom domain -> click on next.

- Select your plan that you need. (you can choose free plan)-> click next. - Configure DNS records like shown in the below picture or you can simple modify the following (dns-text-records.txt) and import in cloudflare DNS manager (you need to modify custom domain, github username and txt record from the above step).

Copy

A records
185.199.111.153
185.199.110.153
185.199.109.153
185.199.108.153

AAAA records
2606:50c0:8003::153
2606:50c0:8002::153
2606:50c0:8001::153
2606:50c0:8000::153

dns-text-records.txt

Copy

<your-custom-domain>	3600	IN	SOA	<your-custom-domain> root.<your-custom-domain> 2041457699 7200 3600 86400 3600

;; A Records
<your-custom-domain>.	1	IN	A	185.199.111.153
<your-custom-domain>.	1	IN	A	185.199.110.153
<your-custom-domain>.	1	IN	A	185.199.109.153
<your-custom-domain>.	1	IN	A	185.199.108.153

;; AAAA Records
<your-custom-domain>.	1	IN	AAAA	2606:50c0:8003::153
<your-custom-domain>.	1	IN	AAAA	2606:50c0:8002::153
<your-custom-domain>.	1	IN	AAAA	2606:50c0:8001::153
<your-custom-domain>.	1	IN	AAAA	2606:50c0:8000::153

;; CNAME Records
www.<your-custom-domain>.	1	IN	CNAME	<your-github-username>.github.io.

;; TXT Records
_github-pages-challenge-<your-github-username>.<your-custom-domain>.	1	IN	TXT	"<text-record-from-github-domains>"

Copy the name servers in cloudflare and Configure it in your domain provider. (For example I am using godaddy for my site). Once you configure this DNS will be manged by cloudflare. This is critical since all the above configuration won't work if you miss this step.

This will take few hours to take effect.

once the nameserver changes got propagated, you can verify the same in cloudflare console. once you verified, the site will be active in cloud flare. till now we have configured the domain provider and cloudflare.

- Now, head back to github and verify the domain in your profile settings. If not able to do this, wait for sometime for the cloudflare changes to propagate. - Once you verified the domain in github, configure the same in github repository settings under custom domains settings. - Congratulations, now your github site will be serving with your custom domain. you can enable HTTPS settings in cloudflare and github repository settings to serve HTTPS by default.

Local aliases using zen aliases

Fri, 05 Feb 2021 19:22:59 GMT

We can configure aliases very easily in bash or zsh. but those aliases are global to all directory. sometimes we need aliases or commands only for some directories alone. this plugin makes the process easy. those aliases will be available only to that particular directory. aliases will be loaded when you cd into the directory and unloaded once you leave the directory.

Configure in zsh

clone plugin from github

Copy

git clone https://github.com/Gokuldroid/zen-alias ~/.oh-my-zsh/plugins/zen-alias

check if the plugin is cloned correctly

Copy

ls ~/.oh-my-zsh/plugins

open .zshrc

Copy

open ~/.zshrc

Copy

vim ~/.zshrc

add zen-alias like this

Copy

# Which plugins would you like to load?
# Standard plugins can be found in $ZSH/plugins/
# Custom plugins may be added to $ZSH_CUSTOM/plugins/
# Example format: plugins=(rails git textmate ruby lighthouse)
# Add wisely, as too many plugins slow down shell startup.
plugins=(
    git
    zen-alias
)

updating the plugin

Copy

cd <cloned directory> && git pull

Configure in bash

copy the file to root directory or somewhere in your system.

Copy

curl -S https://raw.githubusercontent.com/Gokuldroid/zen-alias/main/zen-alias.plugin.zsh > ~/.zen-alias.bash

source the file in bashrc

Copy

open ~/.bashrc

add this line at the end of the bashrc

Copy

source ~/.zen-alias.bash

Aliases syntax

Create .aliases file in the root directory of the project. add your per project aliases like this. You can find a sample file here.

Copy

yc: yarn w @zen-alias/client
ys: yarn w @zen-alias/server
ycommon: yarn w @zen-alias/common
ybg: yarn w @zen-alias/bg-jobs
yb: yarn build
yw: yarn watch

Tips

Excluding the .aliases file in git tree

incase if you don't want others to use your aliases, you can add it global .gitignore or project .gitignore

Copy

cd ~
touch .gitignore_global
git config --global core.excludesfile ~/.gitignore_global
echo ".aliases" > ~/.gitignore_global

Init command for a directory.

we might want to execute some script as soon as we enter the project. like updating local branch or setting up nvm or rvm version etc. zen-alias provides an easy way for this.

define a zen-alias like this,

Copy

init_zen_dir: nvm use && rvm use

You tried hard and you gave your best! still you failed and it's ok!

Fri, 20 Nov 2020 03:41:38 GMT

I was reading a book called Outliers . that made me retrospect my life and the role of luck in my life. There are a lot of factors for one's success, even with a better personal life with some support from others, one can have an edge in your professional life. There are things we can't control in your life. For example, our birthplace or the time. You and the person who works in a different country doing the same job might be earning and getting more rewards than you. It doesn't mean that you're doing less work and you're not capable. you're just in the wrong place. To be successful, one has to be in the right place at the right time.

As a developer, I want stats or exact numbers, not just the stories. So, I decided to simulate the role of luck in an interview. I assumed luck just plays 5% in the interview selection and wanted to compare with the interview result that's purely dependant on talent.

Simulation process:

1.There will be 500 persons with uniformly distributed random luck and talent on a scale of 0 - 100.

2.There are two scores with/without luck.

Copy

public double score() {
    return getTalent();
}

// assuming luck plays just 5 percent in your life
public double scoreWithLuck() {
    return getTalent() * 0.95 + getLuck() * 0.05;
}

3.I will select 10 percent of total people based on these two scores as a two separate selection group. 4.compare the selection and find out how much luck can make a difference in the final result.

To my surprise, even 5% luck made a difference up to 24% when I run the simulation ten times. To give you some idea, if a hundred people are getting interviewed for ten positions, you might be one of two persons who get rejected because of luck.

But if you do the interview twice, your chance of getting rejected because of luck is 24% of 24% (5%). So, try one more time and prove luck means nothing in your life.

Copy

import java.util.Comparator;
import java.util.List;
import java.util.PriorityQueue;
import java.util.Random;
import java.util.stream.Collectors;

public class Luck {
    public static void main(String[] args) {
        Random random = new Random();
        int size = 500;
        int pickSize = 50;
        PriorityQueue<Person> withoutLuck = new PriorityQueue<>(Comparator.comparingDouble(Person::score).reversed());
        PriorityQueue<Person> withLuck = new PriorityQueue<>(Comparator.comparingDouble(Person::scoreWithLuck).reversed());

        for (int i = 0; i < size; i++) {
            Person person = new Person();
            person.setId(i);
            person.setTalent(random.nextInt(101));
            person.setLuck(random.nextInt(101));
            withLuck.add(person);
            withoutLuck.add(person);
        }

        List<Integer> selectedWithLuck = withLuck.stream().map(Person::getId).limit(pickSize).collect(Collectors.toList());
        List<Integer> selectedWithoutLuck = withoutLuck.stream().map(Person::getId).limit(pickSize).collect(Collectors.toList());
        List<Integer> blindLuck = selectedWithLuck.stream().filter(id -> !selectedWithoutLuck.contains(id)).collect(Collectors.toList());
        System.out.println("with luck");
        System.out.println(selectedWithLuck);
        System.out.println("========================");
        System.out.println("without luck");
        System.out.println(selectedWithoutLuck);
        System.out.println("========================");
        System.out.println("blind selection");
        System.out.println(blindLuck);
        System.out.println("=========================");
        System.out.println("blind luck %");
        System.out.println(((double) blindLuck.size() / (double) pickSize) * 100);
    }


    public static class Person {
        int id;
        double talent;
        double luck;

        public int getId() {
            return id;
        }

        public void setId(int id) {
            this.id = id;
        }

        public double getTalent() {
            return talent;
        }

        public void setTalent(double talent) {
            this.talent = talent;
        }

        public double getLuck() {
            return luck;
        }

        public void setLuck(double luck) {
            this.luck = luck;
        }

        public double score() {
            return getTalent();
        }

        // assuming luck plays just 5 percent in your life
        public double scoreWithLuck() {
            return getTalent() * 0.95 + getLuck() * 0.05;
        }

    }
}

Git autocompletion

Sun, 08 Nov 2020 12:30:34 GMT

Remembering the branches and all the available commands are daunting work in git. what if the commands auto-completes itself like shown in the above gif. we can save a lot of typing. not all of us have a typing speed of 90+ wpm right?.

For this, we need to configure our terminal with git auto-completion snippet. you can download the bash-completion file from here. Download the file from these links if you're using bash, or zsh or shell.

You can download with simple curl command also like this,

Copy

curl https://raw.githubusercontent.com/git/git/master/contrib/completion/git-completion.bash -o ~/.git-completion.bash

and add the downloaded file in ~/.bashrc or ~/.zshrc depends on your terminal like this.

Copy

# bash
if [ -f ~/.git-completion.bash ]; then
  . ~/.git-completion.bash
fi

Copy

#zsh
source ~/.git-completion.bash

restart the terminal and enjoy the git auto-completion.

Evaluating documentation frameworks for personal projects.

Sun, 03 May 2020 06:10:35 GMT

I've been trying to write a documentation page for my project (An android app to control your VLC media player) for a very long time. I decided to use some static site builders to get it done quickly.

I am already using gatsby for my blog. I thought of building my gatsby theme as I am pretty much satisfied by Gatsby. but it takes lots of time.

In this post, I will share my opinions on documentation tools that I have considered for my project.

My requirements are:

Appearance should align with my product theme.
It should be user friendly as I am going to use it for one of my android apps (not just for developers).
It should be SEO friendly.
Now I don't have any complex requirements. but it should be extendable in the future.
It should be hostable in GitHub pages using a new repo other than my personal page repo. I didn't want to pollute my blog with docs logic as I might add few more docs for some other projects.
It shouldn't require much configuration/learning for up and running.
It should be simple.

Gatsby-docs-kit:

Pros:

Based on gatsby.
Easily customizable.
We can host it on GitHub.

Cons:

Too much to do for documentation.
Not SEO friendly. but we can configure using plugins available for gatsby. (I didn't want to do that again for docs).

Docsify:

Pros:

Easy to set up.
Comes with four predefined themes. we can easily create our own themes using this tool.
Configurable using plugins. we can write our plugins too.
We can build it and host it GitHub.

Cons:

Not SEO friendly.

GitBook : (I want to try this in future):

Some sample sites using Gitbook Timber.

Pros:

SEO friendly.
No need to set up in our local machine. It is free for personal use.
It comes with powerful analytics.
Drag and drop UI.
It requires less programming knowledge.

Cons:

Vendor lockdown.
We're limited by the customization they provide.
We can't host it on our own.

Gitdocs:

I stumbled up this framework after seeing dev.to's documentation. It is very easy to configure. I felt some limitations from the beginning itself.

Pros:

SEO friendly.
We can build it and host it GitHub.

Cons:

No theming support as of now.
Not maintained very actively (The last commit was made 2 years back). I feel the development has been stopped.
My primary GitHub page is https://gokuldroid.github.io. If I have another project the public URL will look like this https://gokuldroid.github.io/{project_name}. This setup is not easy with Gitdocs. There are open issues and PR related to this.
Not customizable.

Daux.io

Some sample sites using this framework pixolution , soisy.

Props:

It has multilingual support.
We can build it and host it GitHub.

Cons:

The default theme is not user friendly.
There is no much documentation for customization. but some sites using this framework has a really good look and feel.
Only the homepage and documentation page is available. we can't add custom pages.
Not extendable.

MkDocs:

Pros:

Very popular and actively maintained (10K stars in GitHub)
Lots of predefined themes available.
We can build it and host it GitHub.

Cons:

Based on python might not work very well with other frontend tools.

Docusaurus:

There are two major versions of docusaurus. versions two is still in beta and it doesn't have some of the features of version one. but the road map of docusaurus 2 seemed interesting to me. So I wanted to give it a try for my blog.

Pros:

It is maintained by Facebook.
Easily themeable. It uses infima under the hood (Note: this also still in beta).
It comes with a lot of predefined configurations.
Lots of facebook's open-source projects use this for documentation.
It has lots of flexibility.
We can build it and host it GitHub.

Cons:

Version 2 still in beta (but it is pretty much stable).

Based on the observations, I've decided to use docusaurus beta for my app's documentation. You can take a look at the final site here and source code here. Let me know, what is your personal choice in documentation tools and why.

Show me your projects graveyard!

Wed, 22 Apr 2020 14:25:34 GMT

Some of us might have come across very crazy ideas throughout our programming career. when I started my programming career, I didn't know open source and benefits of it back then. I didn't know the best practices and SDLC etc. I have written some shit code to get the things done. I don't even have source code for some of those projects now.I wanted to develop some of these things just to learn even if some opensource version available out there. still I am doing this. but still those projects remind me about my progress from a newbie to senior software developer at Freshworks.

Thumb rule: No one should be using your projects now including you.

FLAMES

For those who don't know what flames check out this link. I developed a GUI application using java and I had two input boxes and a button. fairly simple application. but it took me a day to build back then. In that simple application also, I faced the null pointer exception for some inputs. I didn't bother to fix it as it worked very well for my name and all of my crush's name. my job is done.

Tools used: Java, netbeans

Calculator

A simple Java GUI application. I got some wired errors due to integer overflow. I didn't know how to solve it, thus my application input was limited to Long.MAX_VALUE. I could have used BigInt to support some really big numbers. But I was limited by the technology of my time (complete reference java :P).

Tools used: Java, eclipse

Syllabus app

My first widely used android app by my college friends. An app containing all my college syllabus. It had tab view and dropdown view and lots of text views. I didn't know about DB and content management system when I developed this app. So I hardcoded all data into android layouts (every subject, every semester had a xml layout). that's shit lots of work. But I got lots of positive reviews. I released my app on Facebook as I didn't have a play store account at that time. It is deeply buried in FB feed. you can search for it if you're bored.

Tools used: Java, eclipse with ADT plugin

TicTocToe

An app to play tic-tac-toe over bluetooth.I did complete testing with lots of phones. surprisingly it worked very well in most of the phone. I released this app also in FB. I changed my email, username everything to Gokuldroid because of craze I had on android.

Tools used: Java, eclipse with ADT plugin

Kurukshetra app

I developed this app for my college tech fest. My first app to reach 2k downloads in play store (P.S: Not my account).

Tools used: Java, Android studio beta

The Hindu news reader

My first app to use a backend system. I built using the RSS feed of the Hindu newspaper website.

Tools used: Java, Android studio beta

Secure bay

Phone lock screen app that captures the face using the front camera if you enter the password wrongly. I wanted to use pattern unlock but I didn't know how to develop a gesture-based app. source code can be found here.

Tools used: Java, Android studio beta

Vision website

I developed this site after watching lots of tutorials on youtube and reading lots of books about PHP, angular js (very old version). It is not a responsive one. it will break if you view in mobile phone. I purchased a Godaddy account with the fund I got from my college. I also used Godaddy account's ssh access as socks proxy to bypass my college wifi restrictions.source code can be found here.

Tools used: PHP, angular js, sublime

Gola: static site generator

I wanted to experiment on a node js project. this is the result of it.source code can be found here.

Tools used: Javascript, nodejs, handlebars, express

Some shit cli tools

cli

easy-cli

Tools used: Typescript, node js

Easy google play

A Gradle plugin to publish app. source code can be found here.

Tools used: Kotlin

OMP DSL

DSL builder to generate XML entries. source code can be found here.

Tools used: Kotlin

Grunner - A vscode extension

source code can be found here.

Tools used: Typescript

One minute tips : Less noise in SQL joins

Tue, 25 Feb 2020 17:29:40 GMT

Assume that we need to join two tables based on some condition. For example,

Copy

SELECT users.id, accounts.name from users, accounts where users.account_id = 1 AND users.account_id = accounts.id

Copy

SELECT users.id, accounts.name from users INNER JOIN accounts on accounts.id = users.account_id where users.account_id = 1

Both queries do the same. we are saving 3 words in the first query (this might help you to save 9 secs for every join you make).

Check the query plan for both of these queries, It will be same. I’ve checked this with MySql and Pgsql. I hope this will work in most of the DBMS systems. Let me know if you have any tricks like this.

A missing cmdline util for GIT - NIT

Sat, 18 Jan 2020 07:52:06 GMT

I have been using VS code and cmdline for git. But I always miss something. Autocompletion is not perfect in cmdline and GIT tower is bit costly. So, I created this to make git cmdline experience better.

Installation:

Copy

npm install -g @lazyloop/nit

Usage:

help:

Copy

 nit -h

Copy

nit <command> -h

checkout branch:

Copy

 nit checkout [-r<remote>]

delete branches:

Copy

 nit delete [-m<multiple>]

stage files to commit:

Copy

 nit add

discard local changes:

Copy

 nit discard

NULL vs Empty

Sat, 14 Sep 2019 18:21:41 GMT

I tried to book a ticket in irctc. while giving CCV of my debit card (*mandatory field), accidentally I gave empty value and proceeded to book a ticket. It threw an error after refreshing the whole page as the validation happened (or something went wrong) in the backend. they might have handled in frontend itself. Because of this, I lost my booking window. They might handle millions of requests per day. but they simply failed at this.

It should be a common-sense for a developer to handle null vs empty. To err is human. but repeating history is not acceptable.

I've used a good number of programming languages (Typescript, Java, Kotlin, Ruby, Javascript, Python, C++, C, PHP). There are days I assumed the user is not dumb enough to give empty values for username or password. I haven't developed a real-world software back then. Handling null and empty is always an art. Some programming languages are good at that. Some of them shout like Kotlin 'Yes! we support null safety in our language'. I am not going to give you the definition of null or empty. Do some research and fight with your colleagues about null vs Empty. You will hear some interesting story. I will throw some snippets from the languages I've used so far. Different languages handle them slightly differently. You will get a better understanding of null vs Empty argument once you know about it.

Let's take java first,

Copy

public void demo(String[] args) {  
	Integer count;  
	int countInner;  
	System.out.println(count);  
	System.out.println(countInner);  
}

This one won't compile. It will throw an error, saying variable not initialized. Let's deceive the complier.

Copy

private Integer count;  
private int countInner;  
public void demo(String[] args) {  
	System.out.println(count);  
	System.out.println(countInner);  
}

Yey!! it got complied. What will be the output of this? Can you guess?.

Copy

null  
0

Why the compiler let us compile successfully this time?. When that is a local variable compiler will know for sure that is not yet assigned to any value. we can assign values outside of this method, so there is a possibility of non-empty value inside this function. Thus java compiler won't prevent us from compiling. What if we didn't assign any values to these variables and use these variables like this,

Copy

private static Integer count;  
public static void main(String[] args) {  
	System.out.println(count.toString());  
}

Oops! Here comes the NullPointerException. Now you have some idea about null problem right?.

Let's move on to Empty,

There is a slight difference between no password was given and an empty password. Take a look at this snippet.

Copy

function validPassword(password) {  
	return password != null;  
}  
  
function validPassword2(password) {  
	return password != undefined;  
}  
  
function validPassword3(password) {  
	return password !== undefined;  
}  
  
console.log(validPassword(''));  
console.log(validPassword(null));  
console.log(validPassword2(''));  
console.log(validPassword2(null));  
console.log(validPassword3(''));  
console.log(validPassword3(null));

the output of this will be,

Copy

true  
false  
true  
false  
true  
true

To get a better understanding of this, take a look at this javascript truth table. What if your password validation just checks against null or undefined, you will be allowing the user to have an empty password.

For, == comparison

For, === comparison

weird isn't it?. (Javascript is always weird ;-))

Some programming languages provide safety against some of these issues.

In Java,

Copy

@NonNull  
public static String longToIp(@NonNull Long ip) {  
    return ((ip >> 24) & 0xFF) + "."  
		  + ((ip >> 16) & 0xFF) + "."  
		  + ((ip >> 8) & 0xFF) + "."  
		  + (ip & 0xFF);   
}

But you need to configure tools like FindBugs to get errors and warnings, Can't be better than this?. In some languages null safety is tied into the language itself. thus provides better support for null safety.

In Kotlin,

Copy

var a: String = "abc"  
a = null // compilation error

More details about kotlin null safety,

In typescript (Javascript with types),

Copy

let name: string;  
name = null; // compilation error

More details about typescript null safety

Did you notice something?. No one provides safety against empty values. Because of this, a lot of frameworks and libraries have utils to check empty value.

While writing code, keep asking one thing, "Should I handle empty state?". Handling null and empty state everywhere is also bad. As a rule of thumb, Don't use optional values in the place of mandatory values. If we strictly need a value, throw an error early on, if it null. Please keep this in mind 'there might be a stupid person like me who never reads the instructions'.

"Should a function return null or empty?" is a whole new argument, let's save that for another day.

Java internals : part - 1

Mon, 17 Jun 2019 18:21:41 GMT

Have you came across this meme?

We could say this as backend developer - public api - source code.

We have a rich set of collections and public API in java. We don't need to know the internals of those collections and API in our daily life. But there is a beauty in the internals of Java API's. Here are some of the things I found in the internals of Java's ArrayList code.I assume that you already about ArrayList and the basics about Java collections.

You can get the complete source code from, ArrayList.java.

Copy

//Line : 189
public class ArrayList<E> extends AbstractList<E>
        implements List<E>, RandomAccess, Cloneable, java.io.Serializable
{

The first thing that caught my eyes is RandomAccess interface. It's just an empty interface without any methods or code. It is used just to denote the underlying collection is randomly accessible (Denotes any element of the collection can be accessed at a constant time). Java doesn't support multiple inheritance. But we we can implement multiple interfaces. This might became handy to check the collection is Randomly accessible with constant time.

Copy

// Line : 137
transient Object[] elementData;

Wow!. ArrayList is nothing but plain array internally. ArrayList is just a wrapper on the array. We can write own simple ArrayList. But that might not be performant. But it is always a good thing to try.

Copy

//Line : 179
public ArrayList(Collection<? extends E> c) {
    elementData = c.toArray();
    if ((size = elementData.length) != 0) {
        // defend against c.toArray (incorrectly) not returning Object[]
        // (see e.g. https://bugs.openjdk.java.net/browse/JDK-6260652)
        if (elementData.getClass() != Object[].class)
            elementData = Arrays.copyOf(elementData, size, Object[].class);
    } else {
        // replace with an empty array.
        this.elementData = EMPTY_ELEMENTDATA;
    }
}

Well! It's not a bug, It's a feature ;-). Go through the bug link mentioned in the code. We can see how they track bugs in JDK.

Some elegant code in this snippet,

Arrays.copyOf - Optimized method to make a copy of an array.

Copy

  if((size = elementData.length) != 0) 
  // We could have wrote this as 
  size = elementData.length
  if(size !=0 )
  // But first one is much elegant. When we look into the byte code of these two snippets, there is subtle difference. You can dig into it if you want to learn more about it.

Copy

// Line: 197
public void trimToSize() {
  modCount++;
  if (size < elementData.length) {
      elementData = (size == 0)
        ? EMPTY_ELEMENTDATA
        : Arrays.copyOf(elementData, size);
  }
}

The first usage of modCount++. we will be seeing this snippet more often in this source code. There is a reason for incrementing this variable. If you know the reason already, kudos to you ;-). If you don't know this already, search for the similar lines in the source code. There is a similarity among the methods which has this line. I will tell you why at the end.

Copy

// Line : 287
int indexOfRange(Object o, int start, int end) {
    Object[] es = elementData;
    if (o == null) {
        for (int i = start; i < end; i++) {
            if (es[i] == null) {
                return i;
            }
        }
    } else {
        for (int i = start; i < end; i++) {
            if (o.equals(es[i])) {
                return i;
            }
        }
    }
    return -1;
}

Why they are storing elementData in local variable? any guesses?. One thing I can find is, local variable lookup is faster than class variable lookup. Might be there are some other reasons.

o.equals(es[i]), es[1].equals(o). Is it equal?. Why they didn't use the later one?. Any thoughts?.

For example, we can have something like this,

Copy

ArrayList<Demo> vals = new ArrayList<>();
vals.put(new Demo()); //You know this already.
vals.put(new DemoChild()); // You might know this as well already. (DemoChild is inherited from Demo).

When we have a custom equals, Implementation, we might be in the situation that, demo.equals(demoChild) is true and demoChild.equals(demo) is false. They let us decide on the implementation details if we have custom equal function.

Here comes the prodigal son.

Copy

private void checkForComodification(final int expectedModCount) {
    if (modCount != expectedModCount) {
        throw new ConcurrentModificationException();
    }
}

I asked about modCount++ before right. All the methods increments modCount that mutate the ArrayList in some manner. In a multi-threaded environment, to ensure all the methods are reentrant, we check for modCount in some places.

For example,

Copy

public int hashCode() {
    int expectedModCount = modCount;
    int hash = hashCodeRange(0, size);
    checkForComodification(expectedModCount);
    return hash;
}

hashCodeRange calculation is a long process, By that time some other thread might modify the content of the ArrayList. To ensure this method doesn't return invalid hasCode, at the end of long computation, we are checking for modCount is same as expectedModCount. If that's different the content is modified by another thread and the result that we computed is invalid. In such cases, this will throw ConcurrentModificationException.

Comment below if you find this useful or requires any modification in the article :-).

Why you should care about semantic versioning (SemVer).

Sun, 16 Jun 2019 15:04:20 GMT

Dependency management is always hell.I was a bully once in updating my android app dependencies. Until that dark day came into the picture. Suddenly my app uninstallation events went up. Firebase crash reports waking me up now and then. I don't have billions of users. But still, I have some loyal users (I have around 2L. Not bad for some app developed for my personal use. If you are VLC user you can give it a shot - VLC Stream and remote).

The update was very simple. I just wrapped my code in the try catch block to avoid crash ( I am yet to find the perfect solution for that ;-)) and updated some dependencies. that's it. The culprit was one of the support dependency I updated before. It broke the functionality of splash screen.without it, everything is meaningless in my app. As an indie developer, I don't have much time to test our pet projects and I don't have extensive test cases in my project.

After that I have always thought, 'Why should I change something that isn't broken yet?'. I was reluctant to change to support library dependencies until I had no other go. Because it always broke something in android. Android versioning was not great until they announce this.

The same applies to node projects, ember, react, angular, etc. while choosing dependencies, check whether it follows SemVer. Otherwise, make note of it in package.json (if you are in node projects) or build.gradle (if you are in android projects).So that you can be careful while updating those dependencies. To make our life easier, there are a set of rules that makes this process easier. semantic versioning (SemVer) is something that we can follow in our app releases too.

SemVer is defined as major.minor.patch-(alpha/beta).As simple as that.

Ex, (1.2.14, 1.2.14-alpha).

A change in major version denotes it has some breaking changes. The old API is incompatible with our old code. (Ex 1.20.0 -> 2.0.0)
A change in minor version denotes it has some new functionalities and might have some bug fixes that are compatible (hopefully) with old dependency version. (Ex 1.2.0 -> 1.3.0)
A change in patch version denotes it has some bug fixes and there is no API or functionality changes. (Ex 1.2.2 -> 1.2.3)
(Optional) prerelease tags (alpha/beta) - Added for our own convenience. varies dependencies to dependencies.

Sample projects which follows semantic versioning,

Ember.js, React.js, More about react versioning policy.

Some important rules of SemVer. Go through this. You will understand the four points mentioned above.

Each element MUST increase numerically (Non-Zero and positive number). For instance: 1.9.0 -> 1.10.0 -> 1.11.0
Once a versioned package has been released, the contents of that version MUST NOT be modified. Any modifications MUST be released as a new version. (Once the code is tested with some version, It is always the same forever).
prerelease tags gets lower precedence. ex, (1.0.0-alpha < 1.0.0)
When a major version is bumped. minor and patch versions are reset back to zero. (Ex, 1.23.4 -> 2.0.0).

For more details, check here.

Better cmdline

Sun, 09 Jun 2019 12:25:19 GMT

Every programmer searches for some files more than he breathes in his life. Sometimes it is very daunting to navigate deeply nested folder just to run a single test file. Do you think 'Up arrow never works for me'?. Do you think 'My life is better than my bash history'?. If that so, this article is for you. Sometimes I envy people who use cmd line very efficiently. In this article, I going to show you how my cmd line journey evolved over the period (still I have lots of things to master).

Does native terminal really work for you?

Windows just upgraded their terminal after many many years. MAC's native terminal is good. Is it good enough?.Try to find alternates that suits your need. I have used the following alternate terminals, cmder - windows terminal replacement. iTerm2 - mac os terminal replacement.

Cmd aliases comes to rescue

Clearing cache always fixes the issue right ;-) ?.

Copy

cd /usr/project/master/gokuldroid 
&& git pull --rebase 
&& rm -rf node_modules/ 
&& rm -rf tmp/ && npm install 
&& bower install 
&& ember s

This magic cmd fixes all the issues for me. Over the period we add more and more cleanup methods to this command and this is difficult to remember. What if we could simply do clean_build instead.

To add cmdline alias execute the cmd nano ~/.bash_profile or code ~/.zshrc (For vs code) or subl ~/.zshrc (For sublime users) and add alias like below,

Copy

alias name='cmd'

Adding alias is up to your imagination. you can add anything you want. Some really useful alias.

Copy


alias clean_build='cd /usr/project/master/gokuldroid && git pull --rebase && rm -rf node_modules/ && rm -rf tmp/ && npm install && bower install && ember s'

alias .=cd ..

alias ..=cd ../..

alias ...=cd ../../..

alias ga='git add'

alias gaa='git add .'

alias gaaa='git add --all'

alias gau='git add --update'

alias gb='git branch'

alias gbd='git branch --delete '

alias gc='git commit'

alias gco='git checkout'

alias gcob='git checkout -b'

alias gcom='git checkout master'

alias gcos='git checkout staging'

alias gcod='git checkout develop'

alias gd='git diff'

alias gda='git diff HEAD'

alias glg='git log --graph --oneline --decorate --all'

alias gld='git log --pretty=format:"%h %ad %s" --date=short --all'

alias gp='git pull'

alias gpr='git pull --rebase'

#####find and select like a pro

For this you need to install two plugins.

1.File path finder by facebook

This plugin parses file paths in any output and provides selectable list. For example if you want to view a modified file in current repository just use

Copy

git status | fpp

2.Cmdline fuzzy finder

This video explains the power of fuzzy finder.

(Note: Works at it's best if you configured default apps for the all the file types)

Oh My Zsh

Zsh also called Z shell. It comes with a lot of features that native shell don't have (themes, plugins,spelling correction, approximate completion etc..). You can find installation docs here.

Android loggers analyzed

Sat, 20 Apr 2019 10:57:14 GMT

At first you need to do some changes in logcat console. On the right top drop select Show only selected applications. This will filterout only the application we are working on it.

All the code samples are written in kotlin. Source code is available at Github

Native android logging:

Android has built in methods to do logging.

Example,

Copy

 val TAG = "Demo"
private fun sampleLog() {
        Log.d(TAG, "Hello log")
}

Simple! Is n't it?. It works well for small applications. As you can see in that picture there are lot of unnecessary logs. We can filterout by the Tag name (i.e, Demo in our case).Just type the tag name in search bar.It will show only logs with that tagname.

Cons:

1.You need to define tags each time or you have to write a helper boilerplate which adds tags automatically.I don't want to do this for all my applications , neither you.

2.There is no string interpolation. (It is possible in kotlin, but in java you have to do it manually).

Timber from jake wharton (Github link)

Example,

In MainApplicaiton

Copy

class MainApplication : Application() {
    override fun onCreate() {
        super.onCreate()
        if (BuildConfig.DEBUG) {
            Timber.plant(Timber.DebugTree())
        }else{
            Timber.plant(CrashlyticsTree())
        }
   }
    
    inner class CrashlyticsTree : Timber.Tree() {

        override fun log(priority: Int,tag: String?,message: String, t: Throwable?) {
            if (priority == Log.VERBOSE || priority == Log.DEBUG || priority == Log.INFO) {
                return
            }
            //Send to crash analytics here
        }
    } 
}

Copy

private fun sampleLogTimber(){
        Timber.d("Hello world from %s", "Gokul")
 }

This will produce log like this.

Pros:

1.We can configure the logger by the build configuration easily. All we need is to give custom Timber.Tree. Crash analytics is much easier compared to android native logger.

2.No need to define tags again and again. Timber will automatically all the call site class name in log.We can define tags for one log if we want to.

3.String interpolation is possible and much better stack trace compared to native logger.

4.Rock solid base for all kind of loggers.We can extend it like the way we want or we can integrate it with any other logger easily.

Cons :

1.No Provision to logging to a file.(We can achieve via custom timver tree).

2.No log formatting provision.

3.Adding custom tag is painful one.

Logger from orhanobut (Github link)

It is one of the advanced logger. It supports a lot of things like formatting, exporting logs to external storage and lot of customizations are available.

Example,

Copy

//init logger
Logger.addLogAdapter(new AndroidLogAdapter());

Logger.d("debug");
Logger.e("error");
Logger.w("warning");
Logger.v("verbose");
Logger.i("information");
Logger.wtf("What a Terrible Failure");

//Customizations

FormatStrategy formatStrategy = PrettyFormatStrategy.newBuilder()
  .showThreadInfo(false)  // (Optional) Whether to show thread info or not. Default true
  .methodCount(0)         // (Optional) How many method line to show. Default 2
  .methodOffset(7)        // (Optional) Hides internal method calls up to offset. Default 5
  .logStrategy(customLog) // (Optional) Changes the log strategy to print out. Default LogCat
  .tag("My custom tag")   // (Optional) Global tag for every log. Default PRETTY_LOGGER
  .build();
Logger.addLogAdapter(new AndroidLogAdapter(formatStrategy));

This will produce output like this,

Pros :

1.We can add custom tags.

2.We can jump to callsite from the console it self.

3.We can write our own format easily.

4.It provides all the things that timber has.

5.Output is nicely formatted. No need to do much work to get the sample out like given in the picture.

6.It supports external storage logging.

7.Good documentaion.

Cons:

1.It adds two lines extra for the dashes for every log. Once you get used to that, it will be easier.

There are plenty of loggers available in github.Certain loggers are specifically desinded for some purpose.These are general purpose loggers.

Here is some honurable mentions,

XLog - Method logger.

Debug - Has better IDE support.

Galgo - On screen logger.

Let me know the loggers which you use for android in the comment section below.

Java : Perf matters - 1

Sat, 20 Apr 2019 10:52:53 GMT

Doing microbenchmark is really hard.benchmarking code in microseconds, nanoseconds requires a lot of things to consider.Benchmark results may vary for each computer.Sometimes it differs in the same machine if you do it multiple times as the machine will be at same state ever.To keep the external noise minimum,it is advised to close all the running application so that it doesn't wait for another process to release resources.JIT compiler is playing major role optimising java code on the fly.It does some really great work but in our case, it is also a great factor to consider.For example, when you execute a function multiple times JIT optimises the code. It removes dead code (Empty methods etc, optimising bytecode, inlining functions etc). So we have to fake the JIT compiler so that it doesn't optimise our benchmarking code.In this tutorial, I am going to use JMH Gradle plugin to make tests.

Source code of this tutorial can be found at : PrefMatters

How to run :

Check out code cd dir and run cmd gradlew.bat clean && gradlew.bat jmh

We will start with well known collection arraylist.

Let's prepare a state class that we will use accross each test.

Copy

public static class Constants
{
	private static final int NO_OF_VALS = 500000;
	private static final ArrayList<Integer> vals = new ArrayList<>(NO_OF_VALS);

	public Constants()
	{
		if (vals.isEmpty())
		{
			Random random = new Random();
			for (int i = 0; i < NO_OF_VALS; i++)
			{
				vals.add(i + random.nextInt(Integer.MAX_VALUE / 2));
			}
		}
	}
}

### **Iterating over arraylist :**

1.using foreach

Copy

public void forEachList(Constants constants, Blackhole blackhole)
{
	ArrayList<Integer> vals = constants.vals;
	for (Integer val : vals)
	{
		blackhole.consume(val);
	}
}

blockhole is to consume the val. If we left the loop empty JIT will optimise the loop. That we do't need in our case

2.using indexed loop

Copy

public void forEachIndexed(Constants constants, Blackhole blackhole)
{
	ArrayList<Integer> vals = constants.vals;
	for (int i = 0; i < vals.size(); i++)
	{
		blackhole.consume(vals.get(i));
	}
}

3.using fast indexed loop(intellij IDE says so , we will find out that soon)

Copy

public void forEachFastIndexed(Constants constants, Blackhole blackhole)
{
	ArrayList<Integer> vals = constants.vals;
	for (int i = 0, size = vals.size(); i < size; i++)
	{
		blackhole.consume(vals.get(i));
	}
}

4.using iterator

Copy

public void iterator(Constants constants, Blackhole blackhole)
{
	Iterator<Integer> itr = constants.vals.iterator();
	while (itr.hasNext())
	{
		blackhole.consume(itr.next());
	}
}

5.using list iterator

Copy

public void listIterator(Constants constants, Blackhole blackhole)
{
	ListIterator<Integer> itr = constants.vals.listIterator();
	while (itr.hasNext())
	{
		blackhole.consume(itr.next());
	}
}

Benchmark results:

Throughput method (Higher is better):

Benchmark	Mode	Score	Error	Unit
forEachFastIndexed	thrpt	372.821	± 6.424	ops/s
forEachIndexed	thrpt	409.919	± 14.473	ops/s
forEachList	thrpt	302.609	± 6.424	ops/s
iterator	thrpt	343.567	± 13.681	ops/s
listIterator	thrpt	338.176	± 12.085	ops/s

Singleshot method (Lower is better):

For 5L entries

Benchmark	Mode	Score	Error	Unit
forEachFastIndexed	ss	3.907	± 0.692	ms/op
forEachIndexed	ss	5.299	± 0.890	ms/op
forEachList	ss	5.501	± 1.898	ms/op
iterator	ss	5.053	± 1.113	ms/op
listIterator	ss	6.767	± 1.569	ms/op

For 50 entries

Benchmark	Mode	Score	Error	Unit
forEachFastIndexed	ss	0.013	± 0.002	ms/op
forEachIndexed	ss	0.021	± 0.020	ms/op
forEachList	ss	0.025	± 0.002	ms/op
iterator	ss	0.024	± 0.003	ms/op
listIterator	ss	0.029	± 0.025	ms/op

Conclusion :

forEachList , iterator , listIterator yields same performance most of the time. forEach complies to iterator pattern when we compile and the working of those list internally same. indexed for loop yields better performance than iterators method.Fast index method is really means something in single invoke.Iterator implementation additionally checks for modification,makes method calls internally.In our test we used 5L objects. For smaller size result may be different.Play with this size you will get to know better.I will tend to use fast indexed loop method as my IDE does the auto-completion. This is ok for running app in PC. In android these benchmarks is really important. because we need to squeeze every cycle of CPU.

Adding elements to arraylist :

1.Prepare size and add

Copy

public void preparedAdd(Constants constants, Blackhole blackhole)
{
	ArrayList<Integer> vals = new ArrayList<>(Constants.NO_OF_VALS);
	for (int i = 0; i < Constants.NO_OF_VALS; i++)
	{
		vals.add(i);
	}
	blackhole.consume(vals);
}

2.Just add

Copy

public void add(Constants constants, Blackhole blackhole)
	{
		ArrayList<Integer> vals = new ArrayList<>();
		for (int i = 0; i < Constants.NO_OF_VALS; i++)
		{
			vals.add(i);
		}
		blackhole.consume(vals);
	}

Benchmark results:

Throughput method (Higher is better):

Benchmark	Mode	Score	Error	Unit
add	thrpt	180.190	± 15.306	ops/s
preparedAdd	thrpt	257.963	± 26.253	ops/s

Conclusion :

It is clear that preparing size before adding yields better performance while adding. Initally arraylist will have default capacity of 10. When the arraylist grows internal array is grown to accommodate the data.If we know the size before inserting data it is better to create array with size make ensureCapacity(size) to pre-allocate size so that we can avoid unnecessary operations like creating array,copying old array to new array etc.