Mariano Simone

Single Sign-On (SSO) Should. Not. Cost. Extra. Wrote an article about the boogeyman that SSO has become these days: https://lnkd.in/gDNZ2yXi

29

3 Comments

Mastering Shadow DOM interactions? Discover how Playwright simplifies automating nested elements and unblocks the secrets of Shadow DOM testing, ensuring precise targeting and reducing test flakiness effortlessly with advanced superpowers.  Explore our blog on leveraging Playwright for effective web testing with Shadow DOM at Nagarro! Stay tuned for more in this series from Bharat Garg and myself, on using Playwright to solve intricate challenges. Sincere thanks to Khimanand Upreti, Vineet Bhatt, Alok Sharma, Debaghna Mukherjee, for giving invaluable feedback in refining the content. #WebTesting #Playwright #ShadowDOM. https://lnkd.in/gSwar-bV

38

11 Comments

If I were starting over today building a new startup I'd just build as much as possible on Postgres + Extensions, Docker containers, Amazon Web Services (AWS) AppRunner, and RabbitMQ, and I'd use a well supported Web framework with batteries included over a build your own minimal one. I'd avoid serverless functions, other than for very specific, non-core business things, or using in the short term only if using backend as a service with integrated deployment. I'd lean heavily on AWS for deployment and scaling but only use services that have fully compatible Open Source equivalents, less due to vendor lock in and more due to ability to spin up the whole platform using a Docker compose file on local easily. I'd go with a mono-repo and avoid micro-repos but not be open to whatever size deployable services make sense. If my business was the right shape I'd seriously consider Supabase in order to avoid having to create a whole API for data access and then build a separate API that sits on top to do bigger business logic bits.

2

2 Comments

An interesting take on org design: build lopsided org charts. Start with one monolithic team that is “the product” and only cleave off secondary teams when needed, keeping them small by comparison.

12

4 Comments

Next level: 1. Use the feature flags directly in your tests as well as your code. 2. When you run your test suite, run *every possible permutation* of your flag statuses. Why? Because any flag could be on or off at any time in production, and you need to know how they interact with each other. Yes, this makes your test suite take exponentially longer to run for each feature flag you have. But the reality is that is representing the complexity of your system. Rather than not testing valid code paths, concentrate on reducing the number of available feature flags.

7

The vast majority of content on software engineering I see out there is tailored towards teams that are already high-performing. This makes sense for a few different reasons: high-performers generally write these articles so they write what they know, it's fun and exciting to talk about the latest and greatest, etc. However most teams have a ton of low-hanging fruit that they can address to dramatically improve team performance with low effort lifts. It's very hard to tune an F1 car to go just a bit faster but the simple fact is that most engineering teams are like a commuter car... just by taking a few minutes to put air in the tires you can see dramatic handling and mileage improvements on the daily commute. What are some easier goals to aim for to improve a struggling engineering team? In no particular order: * How big is the average MR/PR? In general a 1,000 line MR/PR isn't four times harder to review than a 250 line MR/PR, it's much much harder. Several thousand line MRs/PRs dropped every few weeks across a team cause lots of issues including merge conflicts and taking a long time to review. After discussing with Product, can tickets be smaller so that merges aren't so big? * How long on average does an MR/PR stay outstanding before getting merged? The older an MR/PR is at merge time the higher the chance of merge conflicts and merge conflicts are a way to generate work but deliver no value to the end user. MRs/PRs that take a long time to merge could be indicative of the team not correctly accounting for the time it takes to review. Should tickets have bigger estimates? * If you're running sprints (and if you're on an underperforming team I recommend switching to sprints ASAP as they bring into focus these types of issues), how often do you have carry-over and how big is it? Improving these estimates is not just an exercise in accounting. A team that estimates work well is one that is not rushed, performant, reviews well, and has time to think about the highest priority tech debt. But most importantly a team that estimates work well is one that delivers on their promises. You may be on a team where these goals have already been achieved or never were an issue in the first place. That's great! But if your engineering team has been underperforming these could be some of the best places to first look. What's some low hanging fruit that you've seen on engineering teams that when addressed had the biggest impact?

11

3 Comments

I believe that when people lament about tech debt, it's a misdirected complaint. Engineering teams will always be producing some amount of technical "debt." Even if we write beautiful, well-tested code that fits some noble architectural paradigm and meets every standard we can come up with, it is guaranteed to be deficient in some way--either now or to the team's (or our own personal) future views on what constitutes good code. When we write code, we come to some consensus as a team around what approaches we generally accept and what trade-offs we are generally willing to tolerate. And that's completely normal and unavoidable. But when we complain about tech debt, what we are actually complaining about is a deficient process. All software teams need to have dedicated time to revise our work. Far too often, we prioritize feature development over all other things. The end results of that deficient process are, invariably, twofold: a product with shiny new bells and whistles and a sea of spaghetti under the hood; and a team of stressed out engineers who have to work in a messy, frustrating codebase every single day.

19

11 Comments

I recently had a discussion with the founder of a CI delivery platform about the value of fixing flaky tests, and it led to the following observation/question: “Over my career, I’ve been in maybe 50 software development teams as an IC. In my current role, I get exposed to 10,000+. If I picked a random person across my entire sample set, they wouldn’t even know where to start with ‘doing the right thing.’ I see way more disengaged engineers who have a ‘give me a ticket and I’ll give you code’ mindset. How would you go about rebooting that?” It’s a great question. I thought I’d write about how we’ve approached this with my last few companies. It’s easy to say that culture is the answer, but how do you build that culture? How do you get every member of the team to care about doing the right thing — and what is the _right thing_? https://lnkd.in/gtMErgGZ

59

4 Comments

The development cycle https://lnkd.in/ggh5NTfh Posted on October 15, 2024 how to clean up code. run your code to make sure it works while (work_to_do){ make a small change test it commit to git } push to git server on local branch try it on another machine collapse git commits for code review test post for code review How to add a new feature run your code make sure it works while (work_to_do){ write a small test while (test doesn’t pass){ make a small change test it } commit to git } push to git server on local branch try it on another machine collapse git commits for code review test post for code review

2

Daily Rant: How do you know what to build if you don’t have requirements that define acceptance criteria? 🚀 Why Requirements Documentation Should Be Every Team’s Best Friend in Software Development! 🚀 We’ve all been there—kicking off a project, ideas flying around, excitement high. But before we dive in, let’s hit pause for a key step: requirements documentation. 📜 It’s not just “extra paperwork”—it’s the backbone that aligns the whole team and ensures we’re building the right thing. Here’s how a solid requirements doc makes every team shine: 🎨 Design Team: With clear requirements, designers know the exact problems to solve and must-have features. This means fewer redesigns, better user experiences, and time saved! 🧑‍💻 Engineering Team: Requirements docs are like a GPS for engineers—showing precisely what needs to be built and why. They keep devs on track, prevent unnecessary back-and-forth, and lead to faster, more focused development. 📈 Product Managers: Clear requirements give PMs a powerful tool to define scope, track progress, and manage priorities. It’s easier to keep everyone aligned, updated, and avoid scope creep! 🤝 Stakeholders: Requirements docs create a shared vision with stakeholders, aligning expectations from the start. This reduces surprises and builds a feedback loop that’s informed and productive. 🧪 QA & Testing: QA teams can create test cases based on requirements, making testing efficient and effective. It’s the best way to ensure the final product meets every specified need before going live. If requirements documentation isn’t a regular part of your process, now’s the time to change that! 🌟 It’s the foundation for project success, aligning the whole team—from design to dev to stakeholders—and setting you up to deliver something amazing. Let’s make it a standard, not an afterthought! 👏

11

2 Comments

Have you worked at a company or on a project where #TDD (Test Driven Development) made a meaningful impact? Let’s spotlight success stories where TDD truly makes a difference. I'd love to hear your experiences, lessons, or even challenges that led to growth. Share your stories in the comments! #SoftwareDevelopment #Programming #Leadership

3

2 Comments

🚀 Single-Threading for Optimal Performance 🚀 Warning: This is a very engineer-y post. ⚙ 📖 Over the holidays, I dove into the insightful book “Modern Software Engineering” by David Farley. If you haven’t read it yet, put it on your list - seriously. One nugget of wisdom that really stood out to me: Parallel computing might not be the silver bullet we often think it is. 💥 Surprised? I was too. We often default to multithreading to boost speed and scalability, but here’s the catch: it can actually slow program execution down. Locks can eat up a ton of time, thread management overhead can wipe out any speed gains, and worst of all, it turns our software into a non-deterministic beast - making it a nightmare to test and debug. 💡 David pointed to an article by Martin Fowler about LMAX, the world's fastest financial trading platform. They process a whopping 6 million orders per second using just a single thread. Mind-blowing. https://lnkd.in/dTJ7w7dC And interestingly: The very same is happening in hardware currently, too. Intel's newest processors have less and less hyperthreading abilities. 💻 Keep learning, keep challenging.

14

5 Comments

After a year of work my book Building A Startup - A Primer For The Individual Contributor is finally available! https://lnkd.in/erThxvpP If you’re working in or around startups as an IC or technical leader this book is for you. It’s based on 20 years of startup work and will help set you up for success to do more with less, move faster and more correct, and spend more time building features and product people love and less time banging your head at roadblocks and bottlenecks.

115

12 Comments

"Curiosity over tenure" -- use quality telemetry in your traces to make debugging accessible. Empower curious folks to debug, not just the most tenured person on the team/company. Metrics are great... they're cheap, enable high data retention, but they're coarse... too coarse to understand what's happening in your system. If you're solely reliant on metrics, you'll need lots of intuition and tenure to understand what's going on in the system. "Oh... it's the cache" says the hero on the team peering over a graph, and doing some inaccessible pattern matching. It looks cool, but it's not scalable. Adding lots of telemetry to your traces ensures anyone with a curious mind can understand what's going on in your system.

3

Here's my 2c: - idiot mode is great - language is lossy - code (as in, working software) is the best grounding function we have - process stifles people's creativity/gives folks an excuse to hide behind - conversely, lack of process stimulates creativity, ownership and initiative more code, less docs! https://lnkd.in/e5vHrx5m

10

4 Comments

How often do you run into a roadblock with an integration, contact a vendor for support, and they ask you for information that they should already have? With most vendors, I bring up a support issue and the first thing they ask is: 🧑‍💻"Can you send me the request payload you sent to our API?" ❌ "Can you show me the error message our API gave you?" 🗒️"Can you send support the output you saw?" It always bothers me when SaaS support has to ask their customer to tell them something about their own system 🙄 That's why, at Rainforest, we built a log of all API requests, responses, and error codes into our portal. We can see, and enable our platform partners to also see, every interaction that has taken place with our API. It cuts down on a lot of unnecessary work to lay eyes on an issue and answer questions.  It often lets our platform partners troubleshoot integration questions on their own. Our support team doesn’t have to ask questions they should already know the answer to, and our platform partners don’t have to act cool while thinking “why don’t they already have visibility into this?” Transparency creates a better support experience for everyone.  The best support interaction is the one that never needed to happen.

22

2 Comments

In Search and Rescue we have a saying "slow is smooth and smooth is fast". The same mindset can be applied to #SoftwareEngineering. Have urgency and bias towards action, but don't rush. Find the root cause. Fix it once. Fix it right. An extra few cycles now will make you faster in the long run.

2

1 Comment

Join me for an interesting session on #devex and how to think about organizational and engineering metrics.

13

On call used to suck for our engineering team. And debugging production issues was a pain because there’s a whole world of possible root causes across various complex systems 🥴 But now we’ve made it simple enough that even our newest engineering team members can debug production issues without escalations 😄  Now they don't have to: - Sift through months of old chat threads to see how a particular alert was handled before - Leave their chat window to go open 86 browser tabs to track down one data point that MIGHT point to a root cause - Pull in members from other teams when they realize an alert originated from an issue in another system that they know nothing about So, what changed? We built an AI Agent to help out with SRE. Aptible AI integrates into our Slack so that when an issue comes up, our engineering team can use it to: - Link to any relevant Slack threads that have mentioned this problem before - Surface data from real-time logs, metrics, docs, and deployment history directly into Slack (with links to the sources) so that they don’t have to leave the chat window to gather the information they need - Determine where in our infrastructure the problem may have originated by using its understanding of how our systems are all working together so that they don’t need to pull in members from other teams TL;DR: Aptible AI allows you to use your brain power to INTERPRET data rather than FETCH it. Aptible AI used to only be available on Slack. But there were >320 million people using Microsoft Teams in 2024... So as of last week, Teams users can use our AI SRE Agent too 🥹👏 📌 Check out the link to the changelog in the comments to read all about our new Teams integration. If you want to try it for yourself, send me a message or comment on this post, and I’ll reach out with a calendar invite 🤓

14

1 Comment