I’ve been using a virtualized workflow for almost a year now, and it’s been great. My work for clients, projects, and specific interests is partitioned off into separate VMs in Proxmox, and VS Code’s Remote SSH plus easy port forwarding helps a lot in making the whole thing feel local.
However, I recently experienced an internet interruption and felt a downside of this workflow: my VMs and Docker containers were no longer reachable. 🙁
https://github.com/lazaronixon/authentication-zero
Found an interesting authentication gem for Rails. Instead of the usual delegation to a gem (where oftentimes you’re not sure what’s actually happening), this one generates the authentication code so the author is in control of what’s happening.
From the readme:
The purpose of authentication zero is to generate a pre-built authentication system into a rails application (web or api-only) that follows both security and rails best practices. By generating code into the user’s application instead of using a library, the user has complete freedom to modify the authentication system so it works best with their app.
Looks like the Linux sandbox VM is running Debian Bookworm.
I just set up Readeck https://readeck.org/en/ on Proxmox using Proxmox Helper Scripts https://helper-scripts.com/scripts.
Just like WordPress, it’s not that I can’t do the installation myself; it’s more that the scripts are convenient and mostly do the same things I would do anyway. So I might as well take a look at a script someone else wrote (and do a cursory check) and use that instead.
Re: Readeck, it seems like a good idea, and being self-hosted, I can be sure the stuff I save won’t be lost when the hosting service goes away (I’m looking at you, Google Reader).
Another useful plugin I found and am using is Widget Options https://wordpress.org/plugins/widget-options/.
I have the recent posts widget, which looks good on desktop but becomes redundant on tablets and mobile devices, so I wanted to hide it on those devices. Instead of writing custom CSS, this plugin does it for me. 🙂
I just tried generating some static pages with this WordPress installation and deploying them to Cloudflare Pages. While it seemed successful (at least from the publishing point of view), it made it difficult to work with some of the plugins that make things convenient.
The Google Site Kit plugin, especially. For some reason, it refuses to use a custom domain and insists on using the current site’s URL as the domain for connecting with Google Analytics. So that means I can’t really host things locally and then generate static pages from the local site without some janky setup, like having the hosts file point this domain locally.
Moving WordPress and MariaDB to the fast cache drive makes things a bit faster.
I was wondering why working on this WordPress blog (e.g., activating themes or plugins) was super slow. Then I realized I had set the installation and database locations on the slow drives. 😕
So in unRAID, I stopped docker and ran the mover (with the shared set to have the cache as primary storage, array as secondary, and the mover set from array to cache).
I just tried this WordPress plugin called Magic Post Thumbnail https://wordpress.org/plugins/magic-post-thumbnail/, which automatically generates a featured image for a post that doesn’t have one.
I’m also using a plugin called Auto Post Thumbnail https://wordpress.org/plugins/auto-post-thumbnail/, which uses the first image in a post as the featured image if there’s no custom featured image set. However, I also have some posts (such as this one) where I don’t really have a picture or illustration to accompany the post.
If your “product owner” demands the team update them every day and turns the Daily Scrum into a daily update, that’s not Scrum anymore.
The daily Scrum is for the team to coordinate with each other so they can work towards the sprint goal.
If the only time the product owner touches base with the dev team is through the daily “updates” provided to them, they’re not on top of things and not really doing their job.
TIL about this option.
git update-index --skip-worktree .obsidian/workspace.json
I have Obsidian on multiple machines synced via Git, and I’m always wrestling with .obsidian/workspace.json conflicts. Today I thought, “Enough is enough,” and decided to fix this.
Looks like with this option, I can set it so that local changes are persisted even if the remote file has changed.
I use Ubuntu Snap to install Ruby on my development machine (because it’s easy), and today I learned that the gem binaries are installed in ~/.gem/bin.
That means I just need to add this to my PATH to use the binaries. I had to do this to get the syntax_tree gem executable to be found by VS Code.
After getting the GPD Win Max 2 and using it as my everyday machine both for work and personal use, I’ve started following the philosophy of not installing any project dependencies locally and instead leveraging containerization technology (mostly via Docker). I’ve talked about this in a previous post, and you can probably think of this post as an improvement on top of that.
I’ve been successfully using this pattern for both work projects and personal ones, and I thought I’d share it.
I needed to analyze some data about the costs of my client’s current CI system, and CircleCI’s insights dashboard—while impressive—isn’t quite up to the task.
Sometimes you just need the plain old “spreadsheet analysis technique” to make progress.
Here’s a quick Ruby script I wrote to get data from all the workflows for a particular project and write it to a CSV file for easy import into a spreadsheet program.
sudo apt install libaio1 libtinfo5
sudo dpkg -i mysql-community-client_5.7.30-1ubuntu18.04_amd64.deb
$ mysql --version
mysql Ver 14.14 Distrib 5.7.30, for Linux (x86_64) using EditLine wrapper
It’s that time of the year again when I get a larger drive, refresh my operating system, and back up important files.
Ever since I’ve settled on having my home folder always be on a separate drive, it’s been really easy to change operating systems, and even moving to a different (more powerful) computer isn’t as much of a chore. If I need to upgrade storage capacity, I just copy everything in my home folder onto the new drive (either by imaging or using rsync) and then make sure that, when I set up the new OS, the new drive is mounted in /home.
One thing that’s been really convenient for me is the ability to move database data along with my home folder. This has been largely because I’ve been using Docker (and Docker Compose) more and more in my projects.
When a Pixel isn’t a Pixel
Presented at the KL Ruby Brigade.
What I learned while building a Slack bot in Ruby.
Presented at the KL Ruby Brigade.
A gentle introduction to modern client-side web frameworks for Rails developers.
Presented at the KL Ruby Brigade.
Braintree recently released v3 of its JavaScript SDK, offering a lot of improvements. Search results for tutorials on how to integrate Braintree with Rails tend to show older articles, so I decided to write an updated one. :)
This is the talk I presented during the first Ruby Conference in Kuala Lumpur, Malaysia. It’s not a word-for-word transcription, but this is mostly the same material I wrote to prepare for the talk.
The secret to creating cross-platform apps is something you already know, but probably aren’t doing.
Whenever you use a method that you know might throw an exception when it shouldn’t, the rescue-log-rethrow pattern is useful for figuring out what happened while preserving the contracts implied in the code.
TDD’s mantra is the following: Red, Green, Refactor. In other words: write a failing test, write code to make it pass, and then make the code better (with the test as a safety net).
I work with really awesome people who do really amazing work. I came upon this commit recently while reviewing a pull request, and I thought I’d share it with you.
Note that some of the information on the diff presented here has been redacted or removed, due to it being proprietary. But the full idea of the refactoring is present.
While working on some code for the Zuora SOAP API, I got bitten by a strange case of inheritance behavior. It seems this is related to what self is bound to during the execution of the statement. Here it is, distilled to its essentials:
Here’s this week’s linklog:
Producing Documentation for your Rails API
The Making of an Expert
Apprenticeship Patterns
So Good They Can’t Ignore You
Architecture: the Lost Years
We’ve commissioned Dekatku to train some developers here at Mindvalley in JavaScript, and it’s been a great experience so far! Today is Day 2 of the training, and we’ve covered some really deep topics.
I’ve started wearing my whitebelt again, and I thought I’d put up my thoughts and notes here for future reference.
I was working on a feature for upcloudify that would use Slack notifications instead of email notifications. One of the challenges I faced was how to build in flexibility for generating messages. I wanted users to be able to provide their own custom notification message and placeholders for items like the S3 download link.
Opal is a Ruby-to-JavaScript compiler, and OpalBox is a jQuery plugin for easily creating a textarea where you can enter and run Ruby code.
I just added OpalBox to this blog, and it’s been quite interesting. I’ve learned how to resize a textarea based on its content (you need JavaScript to do that because CSS just doesn’t cut it), and I’ve added a shortcode.
Angelmmiguel (the author of OpalBox) has just released 0.1.0, which fixes the issue :)
A Code Climate article gave a number of ways to decompose fat ActiveRecord models, and one of them was Extract Query Objects. I’ve been trying to get familiar with a good standard for implementing query objects, and here’s my take.
I’ll be trying out a new format in these articles. I’ll use a Q&A style similar to Socratic questioning.
I fat-finger my code a lot and produce a lot of typos. I do test-driven development, so it’s not as bad, but what’s annoying is when I typo a hash key and the test fails because of a nil, resulting in a very confusing error message.
I was making a gem to have Google Spreadsheets as the backing store for an ActiveRecord adapter when I came upon this interesting method:
I wanted to make a Rake task that would accept something like 1.month or 1.day as one of its arguments. The immediate tool I reached for was eval, and it worked like a charm. However, Code Climate (the tool we use for static analysis) complains about eval, and understandably so; the use of eval is dangerous and can lead to security vulnerabilities.
I’m using it inside a Rake task that is never exposed to third parties. Anyone malicious enough to run the task with a bad parameter to be evaluated would already have access to the system, and could wreak even greater havoc than by running malicious code through the task. I believe this is an accepted risk scenario, and Crystal (one of my colleagues who was reviewing my code) suggested that, since I had a valid point, I could turn off the Code Climate check for this particular instance.
I thought that was justified, but I also thought it was a slippery slope. I’m sure there are ways to pass in a string and have it dynamically interpreted, without having to expose the system to a security vulnerability.
We were looking at HackerRank to see if it could help us alleviate some of the interview and review load on our senior developers.
Some time ago I wrote about how I set up my machine so I could program hassle-free in Ruby while running Windows. It was more of a reminder to myself than anything, in case I ever needed to set things up the same way again.
Fast forward to 2015. When a colleague saw my workstation and realized I was running Windows, he was quite surprised and asked me how I was doing it. I thought of pointing him to my old blog post, but I felt that:
I’ve improved on the whole setup since, and I thought it’s time to share the whole updated setup so that anyone else interested can benefit.
I had to create an automated report for finance and accounting that would send particular columns and their data in a CSV once a month.
Being the lazy developer that I am, I tried looking for code that already existed. I was planning to wrap that code in a Rake task and use whenever to schedule a cron job to send the report monthly.
While I was able to find pre-existing code that did what finance wanted (I also checked with them that the output showed what they needed), the code itself wasn’t easily convertible to a Rake task.
While working on a feature, I was wondering why my test was returning zero records when I’m quite sure the query was written properly (it was just a very simple query about a date range).
I’ve decided to simplify the deployment process and host the blog at S3 instead.
One of the best things you can do during your hackathon presentation is to have your hack available on a publicly accessible platform. You can definitely just host your app on your local machine for the demo, but it is more powerful and impressive if you can involve everyone else in the room and get them to play with your hack during the demo.
I once posted an answer to a friend’s question: Guys, how to win hackathon, and I thought I should expand on it.
You’ve probably heard of the tick-tock model that Intel is using.
It’s a model where every 18 months, there’s a new focus on the chip architecture: the tick is a shrinking of the chip (which requires retrofitting the manufacturing process to support the smaller size), and the tock is an upgrade of the architecture itself using the current manufacturing process.
Update
This is largely obsolete. I have set up a new process to deploy to Amazon S3, and I recommend it over GitHub Pages.
I’m currently moving my blog posts (and collecting entries from various blogging platforms) into Hugo. In case you’re wondering how I’ve set mine up, here’s a guide to what I’ve done.
Yet again, I’ve ported my blog to yet another blogging platform—Hugo. I feel strongly that I’ll be staying on this platform for quite some time, so I’ve been porting over posts from my other blogs.
It’s quite an interesting blogging platform. It outputs static files, which means hosting options are very cheap with a wide variety of providers.
I was invited to pressent at the KL Ruby Brigade and I thought I shuld mix things around by talking about something not very technical but still very much relevant to the audience.
Since I’m in charge (but saying “in-charge” is a bit of a misnomer since in Mindvalley we don’t really have predefined job descriptions) of the internship program for new tech hires, I figures I should share some of the things I consider when interviewing and hiring people.
A friend asked in my Ruby Facebook group about a problem he was having that I think I’ve encountered before (or will encounter in the future). It’s about having to retrieve data associated with a join table.
Here’s the question:
I have set ActiveRecords models like these
I want to do something like current_user.boards.first.role which I expect to load the membership details of current_user AND the first board. Do you have any idea how can I achieve this? Do you suggest another approach?
Work In Progress
I’m still in the middle of writing this article, and I probably will be always updating this.
I’ve just recently (re)installed Linux on my Acer V5 122p. Had to fiddle around with UEFI settings thinking to preserve my Windows 8.1 partition, but eventually I had to bite the bullet. I wiped the whole partition clean and turned off UEFI so my installation would be a lot simpler.
TL;DR Use Swagger and Grape to create self-documenting APIs that automatically update themselves from metadata in the code. It also provides an API explorer to play around with.
During the AngelHack KL event, I (being the backend developer of the team) had to create an API to facilitate communication between the iPhone app and Twilio. Since it was a hackathon, there was not a lot of time to completely hash out API protocols (as well as explaining to my team mates how RESTful APIs work).
I was able to solve this by using Grape and Swagger. I can develop the API at my own time. When a particular endpoint was finished, I can push to production and my team mate is immediately able to see the documentation. He was also able to make use of the API explorer to test out the API’s responses to his queries.
We were able to quickly develop the whole project quickly and easily because of this, and here’s a quick look at using Swagger with Grape.
Paul Graham’s (of YCombinator fame) essay on the difference between the Maker’s Schedule and the Manager’s Schedule is a timeless piece of advice that is sadly not as well known as it should be. The essay provides an interesting, if not logical reason behind why programmer’s skip meetings called by managers, while managers often tend to interrupt programmers.
Here are some of the important points in his essay:
I’ve done the unthinkable and I have a confession to make.
I’ve switched my development machines to Windows.
Here’s the story of my heresy … and my possible salvation.
One of the most often used and ironically least understood concepts in the Ruby programming language is the block. If you’ve used one of the Enumerable mixed-in objects, you’ve encountered the block syntax.
It’s always easier to deal with a ruby binary that’s installed server-wide rather than a local user installation (done via rvm or rbenv). Server initialization scripts don’t need to do path magic and profile sourcing in order to find the correct binary. Here I how to install ruby-2.0.0-p195 in a way that your system administrator will love you for.
Here I show how to perform a rolling restart using Capistrano, and making sure that the load balancer (in this case Linode) properly reroutes traffic to the still-running servers while the others are being restarted.
WARNING I’m still ironing out a few bugs for the rolling deployments. The issue appears to be that Capistrano executes
deploy:create_symlinkin parallel so the apps still experience downtime before the restart happens. The concept itself is sound but the code is not yet production ready!!
PrivatePub is an awesome gem by the renowned Ryan Bates of RailsCasts fame. It provides an easy way to use Faye as a pub/sub provider. Watch this episode to see it in action.
You don’t have to bundle the gem with a Rails application, though; you can deploy PrivatePub as a standalone Rack application on Heroku. This allows you to run PrivatePub in a web dyno instead of a worker, making maintenance easier (by separating repositories/codebases) and saving on hosting costs as a side effect.
At Mindvalley, employees can have any kind of hardware/software they want, provided that they can prove it is necessary for the work they’re doing or that it increases their productivity. In my case, I opted for the setup shown here (picture courtesy of my colleague):
(The blank monitor on the left was an Ubuntu VM in screensaver mode.)
I just bought a VPS (virtual private server) a couple of days ago, and I’m enjoying tinkering with it. It’s not the most powerful beast, though; it only has 64 MB of RAM and 128 MB of swap. I won’t say the name of the company I’m subscribed to since my box isn’t exactly a bastion of security, but you should be able to discern the shell provider from how I describe my escapades. :D
While mucking around on my computer and generally getting bored, I found a few neat programs to install. They’re really cool, and your Linux computer shouldn’t be without them. :)
My data drive had some filesystem tree corruption, and I was forced to rebuild it. I figured I’d just reinstall Kubuntu too, since I inadvertently installed Ubuntu Desktop, which kind of doubled the number of applications on the menu.
Here are some of the things I did to remind myself in case my system crashes again:
I’ll add to the list as I walk through the rebuilding of my settings.