“The mind behind Linux | Linus Torvalds”

Sometimes you can see a problem in a different way and rewrite it so that a special case goes away and becomes the normal case. And that’s good code.

To me , the sign of good people I really want to work with is that they have good taste.

This is a small example , Good Taste is much bigger than this.

Good taste is about really seeing the big patterns, and kind of instinctively and knowing what’s the right way to do things.

Background

• I prefer rvm over rbenv, maybe I should use rbenv insead ):
  • rvm/rvm: Ruby enVironment Manager (RVM)
  • rbenv/rbenv: Manage your app’s Ruby environment
• rvm list known not showing the latest ruby version
• rvm instal 3.2 fails
• rvm fix-permissions not working for “_rvm_log_dotted:23: permission denied” error
• I noticed these two pieces of information
  • “NOTE: Even though this is all known rubies, RVM can install many more rubies not listed :)”
  • “Because https://cache.ruby-lang.org/pub/ruby/3.2/ has no "Ruby-Preview.tar.gz" file in the directory"
  • which means it should be able to install any ruby version listed in pub/ruby/ with rvm

Steps

1
2
3
4
5
6
7

rvm reinstall "ruby-3.2.0" --with-openssl-dir=`brew --prefix openssl@3` # not ok, [ruby don't support openssl 3 yet.](https://github.com/ruby/openssl/issues/369)

brew uninstall openssl@3
brew reinstall openssl@1.1

# new shell
rvm reinstall "ruby-3.2.0" --with-openssl-dir=`brew --prefix openssl@1.1` --disable-binary # ok

Other ref

• _rvm_log_dotted:23: permission denied: | Tinyfeng’s blog

• TODO learn you a haskell
• OK gopl
• TODO beyond feelings
• TODO sicp
• TODO unix network programming volume 1: the sockets networking api, 3rd edition
• TODO [optional] High-Performance-Browser-Networking
• OK [optional] clean code
• OK algs4 (20230421)
• OK A philosophy of software design(20230421)

• review: unix-processes 15 * 15 min = 225 min(done 20220517)

• read: 元编程 240 页 / (10页/15 min) = 360 min(done 20220521)

• listen/review: 史蒂夫乔布斯传 (76 - 18 + 1) * 20 min = 1180 min(done 20220524)

• read: 暗时间 244 页 / (10页/15 min) = 366 min(done 20220623)

• read: CSAPP 730 页 / (10页/20 min) = 1460 min(ff)(done 20220512 -> 20220925)

• review: 现实一种(余华) (done 20220925 -> 20221001)

• read: unix编程艺术 (done 20200704 -> 20220924)

• read: refactoring (done 20220929 -> 20221012)

• read: DDD (done 20221107 -> 20221213)

• Bi-weekly Tech Sharing

Agenda

• Experience
• Testing
• Refactoring
• References

Experience

• best thing learned this year

• I’m proud that we have testing culture and code review

• how I test my code before?

  • coding -> run program -> postman / repl -> verify by eye -> QA

• how I test my code now?

  • coding -> add tests -> green -> refactoring -> coding -> repeat -> code review

• hopefully I’ll benefit from TDD style sooner

Testing

• raise a question: we’re relying on tests to guarantee code work as expected, who guarantees our tests are correct then?

  • short answer: we can only rely on simple tests
  • too simple to be wrong(ideal world)

• testing proportion

• testing helped on

  • building confidence
  • building solid(bug free) code as possible
  • gives me a button to click
    • “if you ever get that button, you’ll never want to lose it or work other way round so you wouldn’t have it.”

• testing drives me to write simple code

  • I want class to be small
  • I want responsiblilty to be single/simple/clear
  • I want method to be short
  • I want interface to be well defined
  • I love good names

• example of using let(private git repo pull request)

  • demo for independant tests
  • avoid mutable variable
  • discipline posed upon assignment/mutability

• key points:

  • write simple code & test when possible
  • more unit tests
  • “write enough tests to be confident”

Refactoring

• let’s put performance aside while coding & refactoring

• let’s focus on clarity/simplicity/readability/maintainability, not performance

  • make it simple
  • make it easier to read/review
  • make it easier to change/maintain

• How I do code refactoring?

  • clean up feature flags
  • check code coverage and make it up to 100%
  • remove code smell
  • simplify code
  • enjoy the tiny changes/improvements
  • enjoy the green dots
  • enjoy the red dots(it shows I didn’t fully understand my code)
  • remove unused/dead code(unused)
  • check with flog(rely on metric tools, not “feeling”)
  • choose a better name for variables/methods/class/file/moudles
  • simplify tests
  • I got better understanding of the code, and I can repeat the above process…

• example of simplifying code(private git repo pull request)

• The point is: “business complixity is the lower limit of system complixity”

  • so I want my code to be as simple/readable as possible
  • try functional thinking: e.g. use filter/map/reduce
  • ruby: Enumerable module is amazing

• key points:

  • lean on tests, no tests, no refactoring
  • do your refactoring with small steps(highlighted in both “refactoring” and “99 bottles of OOP”)
  • refactoring is just pattern matching(focus on code smell) and follow the solution

References

• book: 99 Bottles of OOP — Sandi Metz
• book: Refactoring
• gitrepo: A tool to identify potentially unused code.
• gitrepo: sandimetz/99bottles_ruby
• guide: Better Specs. Testing Guidelines for Developers.
• paper: algorithm = logic + control.pdf
• post: “You Don’t Know What You Don’t Know”
• post: Let vs. Let! vs. Instance Variables in RSpec
• post: Simplifying Tests by Extracting Side-Effects
• post: 测试指南 - Rei
• quote: “Do yourself a favor and make sure your test coverage is good before you start an upgrade.”
• quote: “Pain in your tests should push you to change the design of objects being tested”
• quote: “if you ever get that button, you’ll never want to lose it or work other way round so you wouldn’t have it.”
• quote: “Running our test table: Click the Test button”
• slide: slide page7: “Focus on Simplicity & Clarity, Otherwise, you don’t fully understand.“
• talk: “Fearless Competence / Conquer the fear with Test.”
• talk: Clean Code - Uncle Bob - YouTube
• tool: Catalog of Refactorings
• tool: Confessions of a Ruby Sadist sudo gem install flog
• wiki: Code Smell
• wiki: Rules Of Optimization
• wiki: Test Driven Development

At last

these are just my limited experience on testing & refactoring; I’m pretty sure I have a lot more to learn

I really want to learn from everyone about these two topics

pls correct me if I’m wrong. pls discuss with me if you’re interested~ Orz

This one doesn’t have a TL;DR

Background

• I learned to write tools for my own from this repo
• I heard career advice of “Spend 30 minutes each day improving something about your process (automate the thing you repeat the most) or environment (editor, shell, OS, etc).”
• I’m using “Oh My Zsh”, and have my own omz plugin, the one causing issue is another private repository

What I did

• I wrote two specific shell aliases for rspec, re for running specs without error backtraces, ree for with error backtraces

  1 2	alias re="DISABLE_SPRING=1 bundle exec rspec --format=progress --no-profile" alias ree="DISABLE_SPRING=1 bundle exec rspec --format=progress --no-profile --backtrace"

• One day, I found it looks better if I change them to this(alignment):

  1 2	alias re=" DISABLE_SPRING=1 bundle exec rspec --format=progress --no-profile" alias ree="DISABLE_SPRING=1 bundle exec rspec --format=progress --no-profile --backtrace"

  

• Later, I forgot the change

• Somehow, I find the frequently used re command never got remembered by type re and press the up arrow(which acts as searching from history in OMZ)

Trouble shooting

• I didn’t think too much about what I’v recently changed
• I thought I screwed up my OMZ environment, so I keep creating new terminal session, didn’t work
• I thought my history file is too big, so I cleared the history file, didn’t work
• I put this issue aside for a few days without any solution, suffering from typing/copy-paste the frequently used rspec commands again and again
• A few days later, I can’t stand it anymore, and at this time, I found that only that rspec related command won’t be remembered by history!(still didn’t connect to previous changes…)
• Until second day when I tried to optimize my OMZ plugin, I noticed the git history: THERE IS A LEADING SPACE IN MY RSPEC ALIAS!!! because of the code alignment

The reason

• I learned the trick about 2 years ago from a youtuber “Engineer Man”: A leading space in command line cause it won’t be written into history
  • https://www.youtube.com/watch?v=Zuwa8zlfXSY&t=384s
  • sadly I can’t google it out the name of this feature…
• To be honest, I almost got no chance to use that trick, only used once time on production server(didn’t mean to hack xD)
• But I remember the tick and it’s this trick and the coincidence that caused my alias not written into history!

The fix

• Remove the leading space in shell alias, simple enough

Learnings

• Be aware of your changes, make sure you fully understand the output
• When unexpected behavior happened, checkout your changes first
  • Maybe “SRE has found that roughly 70% of outages are due to changes in a live system.” applies to my little system too…
• Don’t tolerate the inconvience for too long, find it, solve it
• Perform tests after changes even for my own tools

Original Post

See implementation

Core idea: using multiple repository remote

WIP

最初一直不喜欢rails里的activejob

原因很简单, 日志里它的日志很难懂

后来花时间读了一下文档, 了解到了gid这个概念, 知道了它里面有一些和controller里一致的rescue_from等等功能, 觉得它确实有优势

日志也能看懂了, 不那么排斥它了

甚至因为gid的缘故, 有点喜欢这个设计

但是, sidekiq里提倡的最佳实践是: 使用简单参数, 很清晰易懂, 用gid其实也做到了一样的事, 而且他可以传递model实例, 挺好的优势

直到昨天, 看到同事的一个pr修改了activejob的参数(由model实例改为id), 我提出是不是没必要这样, 因为前面说到的原因

但是他展示给我 如果在log里搜索 “Sidekiq::Extensions::DelayedClass ARGS” 会发现, 虽然perform接口的model实例会被序列化为gid, 但是这种延时的任务, 实际上还是序列化为了yml

所有model实例的所有信息都被存储了下来, 然后做反序列化, 所有信息都被记入了log里面, 太糟糕了

• 反思.1: sidekiq的代码没读全, activejob的代码没读全, 重大的疏漏
• 反思.2: 简单的是好的, 简单的最佳实践也需要认真了解其背后的逻辑
• 反思.3: 不要对自己信奉的东西有太强的自信, 多多沟通和交流, 谦虚的向大家学习

Error.1 NS_ERROR_FAILURE

Using macos to debug c program is annoying…

Tried to use vagrant vm, but I met error while booting the vm

1

VBoxManage: error: Details: code NS_ERROR_FAILURE (0x80004005), component MachineWrap, interface IMachine

Solution

https://stackoverflow.com/questions/52689672/virtualbox-ns-error-failure-0x80004005-macos

I tried reinstall Virtualbox, not working

I tried reboot mbp, this time it worked

Don’t know the root reason yet

Error.2 umount: /mnt: not mounted

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

$ vagrant up

...

The following SSH command responded with a non-zero exit status.
Vagrant assumes that this means the command failed!

umount /mnt

Stdout from the command:



Stderr from the command:

umount: /mnt: not mounted

then I can vagrant ssh into the machine, seem above error is not a big deal

But I want to make it go away…

Reason

seems related to vagrant plugin vagrant-vbguest

1
2

vagrant plugin list
vagrant plugin uninstall vagrant-vbguest

Solution

Unistall the plugin and reboot the vagrant vm

1
2
3
4

vagrant halt
vagrant destroy
vagrant up
vagrant ssh

Error.3 Remote connection disconnect. Retrying and ask for ssh password

https://github.com/puphpet/puphpet/issues/1253#issuecomment-145429092

for some dumb people like me, do not use /home/vagrant as your shared folder (in your vm), because the .ssh files are not accesible.

don’t understand the reason, but this is the cause, remove the config.vm.synced_folder resolved the problem

How to debug

1
2

vagrant ssh-config
vagrant ssh --debug

Solution

rm config.vm.synced_folder in Vagrantfile and try again