Bitbucket Pipelines costs slowly boiled me alive

When we started Design and Test Lab (software outsourcing business) back in 2011, git was our default choice to work with source codes. Initially, we put a server box under a work desk in our office and spent some efforts to set-up it and maintain it. It was some pain to administer it, uptime was not so good. And I was out looking for cloud-based alternative.

It was 2011, GitLab was launched only three years later (in 2014). At that time I've found two competitors: GitHub and Bitbucket.

I don't remember the exact numbers, but I've chosen Bitbucket because of the price per user. As far I remember, GitHub had pretty big pricing (something like $4 vs $20 per user). That's how we ended up with Bitbucket.

Read full article: https://www.ciperf.com/bitbucket-pipelines-costs-slowly-boiled-me-alive.html

My biggest self-host fear. :)

Шановні, запрошую сьогодні о 19:00 до себе на подкаст (наживо). Будемо обговорювати тему поразуміння інтересів бізнесу (топ-менеджменту) та розробників.

В гостях Влад Зубков — macOS/iOS developer @ Nektony, більше 10 років досвіду, у нього є декілька цікавих історій.

Струм буде в LI: https://www.linkedin.com/events/businessvsdevelopers-findingthe7209572701005324290/about/

I've created hourly benchmark for Bitbucket Pipelines. It records: time to push, time to provision a CI server, and a performance test for the CI server itself.

Very first conclusion: Bitbucket Pipelines performance is not a constant. I'll share some details and methodology later. Follow me for updates.

Which CI service to benchmark next? Gitlab? Github?

There are lots of cases when self-hosting is better than hosting in cloud. And self-host is way cheaper and performant than clouds. Here is my short talk about it:

https://www.youtube.com/watch?v=0VynTTiK9g4

Backend Frameworks with the most letters in its name:

1. Ruby on Rails - 11 letters
2. Spring Boot - 10 letters
3. Laravel - 7 letters
4. FastAPI - 7 letters
5. Express - 7 letters
6. Django - 6 letters
7. Meteor - 6 letters
8. Flask - 5 letters
9. Nest - 4 letters
10. Koa - 3 letters

So Ruby on Rails is the best feamework, obviously.

And how do you choose a framework for your app?

softogenic (software + -genic) — adjective, of or pertaining to a process or substance created by software;

A softogenic disaster is one that arises from errors and faults in software.

Please note this new word. You will use it very often soon.

CrowdStrike’s CEO:

Why is Bitbucket Pipelines slow? How to optimize bitbucket-pipelines.yml by minimizing build steps

I'd like to share my experience with optimizing Bitbucket Pipelines performance. I've noticed that in some cases, test setup procedures take some time (several minutes). This includes the installation of needed packages, libraries, etc. In most cases, we don't need to do this every time we run the pipeline. It is possible to pre-build a Docker image with all dependencies and save time and money.

Read full article: https://www.ciperf.com/why-is-bitbucket-pipelines-slow-how-to-optimize-bitbucket-pipelinesyml-by-minimizing-build-steps.html

Is there such a thing as a "bad fire"?

I've experienced many "bad hires," but I can't recall a "bad fire"—a situation where we fired someone and later regretted it. In every case, my only regret was not firing sooner.

What's your experience?

It took me 10 years to understand the most important quality of a team mate.

I’m looking for positive thinking people and deep negativitiy is a huge red flag 🚩 for me.

Here is a brief video about this:

https://www.youtube.com/watch?v=LlcrzKN09s8

I'm 41 today. 🥳

If you have spare $5-10, I'd appreciate if you could donate to my friend Juliia Pashchuk, she is a volunteer in Kharkiv 🇺🇦 and she is going to 350 kids (refugees from Kharkiv region), gathering stuff for new school year on September 1st. The donation link is here: https://send.monobank.ua/jar/7xa6JmgS6a

Thank you all, I love you all! ❤️

Ulyana Obrizan, 3 yo. :)

The Sure Way to Waste Money on Software Development

For over ten years now, I’ve been developing custom software. Our company has worked on small projects for solopreneurs, where we could go from an idea to a finished mobile app in the App Store in just a couple of weeks, as well as on large projects for multinational corporations, where multiple applications were developed by a team of several developers over the course of years. I’ve had over a hundred such projects and clients, and I’ve seen a lot.

My name is Vladimir Obrizan, I'm director at Design and Test Lab and I want to start a series of stories called "How to Waste Money on Software Development."

Let me begin with a story about one project. Through a freelancer platform, I was approached by a school teacher from the U.S. (let’s call him Chris). In his first message, Chris said that he wanted to create a simple mobile app for iPhone to help him organize various events. After hearing his ideas, I told him that it was indeed a simple mobile app and that we could build it in two weeks with one developer.

The app’s logic was really simple. The end user would install this app on their phone. Then, they would grant access to their contact list, select the guests they wanted to invite to their event, enter the event name, date, and a small questionnaire like: “Do you prefer meat or fish? How many friends will you bring? Will you stay overnight?”

The mobile app would generate a message based on a template, insert the guest's name from the contact list, and send an SMS to their phone number. Of course, the iPhone doesn’t allow mass SMS sending, so the user had to send each SMS manually, but we automated the creation of the SMS itself, so pressing the “Send” button 10 times wasn’t much trouble for the user.
We presented the app to the client, and he really liked it. Then he said, “Now I want to collect statistics on the responses and send them to myself as a PDF via email.”

I was a bit stunned. Sending and receiving SMS messages is a basic iPhone capability, but reading those SMS messages programmatically is not possible, even now, 10 years later. That’s just how iPhone security works.

I started to get a little nervous. How could I tell the client that it wasn’t possible to implement this feature using an iPhone app and that he had essentially wasted two weeks and, most importantly, $1,500?

I took a deep breath and said, “Chris, it’s not possible to implement this with an iPhone app. iPhone security rules don’t allow SMS messages to be read programmatically. To implement this feature, we would need to build a web application, connect it to an SMS gateway, purchase a phone number, and process the replies programmatically. My rough estimate is another two weeks and $1,500.”

In this context, it was the right answer — Chris agreed to increase the budget by another $1,500, and even more, because we ended up building an SMS form builder on the website, setting up a sequence of SMS messages (essentially implementing chatbot-like functionality, but over SMS), and creating beautiful PDF reports. We even hosted everything on Google Cloud. Scalability, fault tolerance — everything like in the big leagues.

What did this experience teach me? Not to rush into coding whatever the client asks for without thinking. In the past, if someone asked me to "make a mobile app," I’d make a mobile app. If they asked me to use MongoDB, I’d connect MongoDB. If they wanted the site painted green, I’d paint it green. After all, I wasn’t responsible for the business success that depended on the app, right? But the customer isn't always right when it comes to choosing technology. Sometimes their choice can be fatal for the project: the customer might not be able to integrate the app into their business processes later. In that case, time and money are wasted.

⬇️⬇️⬇️ cont in comments

How a Good Idea Can Become a Monster

I use Thinkific as a learning management system (LMS) for my online course, "Python + Selenium: Test Automation for Web Applications." Thinkific is a SaaS platform, and I pay $150 every month for it. Over the past three years, my total investment in the LMS has been $5,400.

There are times when I don’t have any students, so essentially I’m paying $150 per month for nothing. Shutting down the course isn’t an option, though, because the course is already set up. If I get new sign-ups, I need to have a platform ready for onboarding and delivering the content without the hassle of re-uploading everything.

Over the past year, I’ve become a fan of self-hosted solutions because the cost-to-performance ratio of modern servers is incredible. For instance, I rent a server from Hetzner with 14 Intel i5 cores, 64 GB of RAM, a 1 TB SSD, and unlimited 1 Gbps internet for just $40 per month. You can imagine how many services I can run on such a powerful machine!

My first thought was: "This should be simple—I just need to find an open-source LMS and install it."

I remembered that there’s a popular open-source LMS called OpenEDX. Great! I opened their installation guide, which offered a one-click installation with Docker. Perfect!

I clicked it and…

I discovered that the modern version of OpenEDX requires 12 containers to run! The installation alone took about 25 minutes just to apply the initial migrations and get it up and running.

Then, I spent an hour trying to create a course within the LMS…

That’s when I realized that OpenEDX is overkill for me. It's complex in every way: system architecture, UI, UX—you name it. And frankly, I don’t like it.

In the meantime, I’m working on building my own LMS using Python and Django. Any suggestions for a minimalistic open-source LMS?

Interesting software error

I encounter software errors all the time, they really annoy and frustrate me. But some errors are funny.

For example, Remarkable home page — 404 not found. :) Usually this happens with some old pages forsaken in a web-site deeply. But I've seen the first time when the very root home page is not found. :)

Try yourself: https://remarkable.com/

Also, any suggestion for e-ink note-writting device?

The Sure Way to Waste Money on Software Development: Focus on the Team!

I'm continuing my series on how to waste money on software development. In the previous video, we discussed how to avoid wasting money by sharing your strategic plan for the app with your developer in advance. That way, they can suggest a technological solution that not only solves the current problem but also scales for at least the next year or two, saving you from having to rewrite everything from scratch later.

Today, we’re going to talk about an important topic: choosing a contractor and hiring a team.

So, you’re an entrepreneur with a brilliant idea for a new app, and you’ve decided to hire a developer. Do you consider yourself an experienced client? Or is this your first time ordering development?

An experienced client has gone through the hiring process several times, understands technologies, the roles in a team, and the development process, and can distinguish a good team from a bad one. Inexperienced clients often associate software development with coding. That is, they think someone who can program will be able to develop the app on their own.

My series is called "The Sure Way to Waste Money on Software Development," where the title implies that the client loses money. But that’s not entirely true. As a business owner, as someone who has executed software projects, I’ve lost money too! And now I’ll tell you one story.

It was in the early 2010s. I found projects on a freelancer platform and handed them off to developers from my company. I only accompanied the project up to the deal stage, when the budget was agreed upon. Once the deal was made, the project was handed over to the developer. The developer was responsible for communicating with the client, completing tasks, and showing results.

The client found a developer—doesn't that sound like the recipe for success for my business and the client’s startup? No! I thought this model was viable and would work too.

But I was wrong!

Cont. in comments ⬇️⬇️⬇️