KTH
Lecture Participation Tracker 2025
Here we track active participation in lectures.
To do this, you record as a comment the question you make to presentations or demos during the lectures.
Also, provide the title of the presentation/demo.
Lecture Participation Stats (Updated on 2025-10-08 14:39:48)
| Index | Student Name | Number of Lectures Attended | Lecture(s) attended |
|---|---|---|---|
| 1 | adrianthees | 6 | L2 L3 L4 L5 L6 L7 |
| 2 | anica279p | 6 | L2 L3 L4 L5 L6 L7 |
| 3 | alexmatcov | 6 | L2 L3 L4 L5 L6 L7 |
| 4 | putastep | 6 | L2 L3 L4 L5 L6 L7 |
| 5 | AlessandroColi | 6 | L2 L3 L4 L5 L6 L7 |
| 6 | THEosusi | 6 | L2 L3 L4 L5 L6 L7 |
| 7 | ujanjan | 6 | L2 L3 L4 L5 L6 L7 |
| 8 | riccacocco | 5 | L2 L3 L4 L5 L6 |
| 9 | hexakon | 5 | L2 L3 L4 L5 L6 |
| 10 | fianmian | 5 | L2 L3 L4 L5 L6 |
| 11 | ShreyasSawai09 | 5 | L2 L3 L4 L5 L6 |
| 12 | adammeh | 5 | L2 L3 L4 L5 L6 |
| 13 | AhmadAddee | 5 | L2 L3 L4 L5 L6 |
| 14 | peschwartz | 5 | L2 L3 L4 L5 L6 |
| 15 | somaiyaabdulrahman | 5 | L3 L4 L5 L6 L7 |
| 16 | foodelevator | 4 | L2 L3 L4 L5 |
| 17 | N-Asadnajafi | 4 | L3 L4 L5 L6 |
| 18 | edv1n | 2 | L2 L3 |
| 19 | dania-sami | 2 | L5 L7 |
| 20 | VincentLagerros | 1 | L2 |
| 21 | VolcanoCookies | 1 | L2 |
| 22 | SaiBon99 | 1 | L2 |
| 23 | Herdi2 | 1 | L2 |
| 24 | aldenbro | 1 | L2 |
Setting up a CI pipeline with GitHub Webhooks and REST API feedback - Demo Q: Why did you choose the CI to run on every push instead of a PR to the main? A: For simplicity purposes. In a real world scenario you would ideally run the tests and compile before you merge to the main branch to avoid bad code in the important branches.
Question related to: https://github.com/KTH/devops-course/pull/2704 - Demo Question: Is there a way to make the latency shorter, due to the cloning and installing of dependencies? Answer: Yes, once you install your dependencies, they should be cached, making the process faster and using incremental builds rather than rebuilding from scratch.
Question related to: Week 2 Demo Github Actions #2700 Q: Are the rules implemented by cubic ai automatically imported in the GitHub repository or the user have to do specific action for implementing it? A: after logging in in cubic AI with the GitHub account you chose the repository that you want, and GitHub automatically set the new rules on the project.
Question related to: #2700 - Cubic Demo Q: Is this service a free or is there a paid version? Moreover, can be hosted locally with your own LLM? A: No, this is a paid service, but codex is an alternative.
Question about #2700 Demo Q: Advantages of using an AI that creates pull requests like Cubic, instead of a "local" LLM like Cursor or Copilot? A: Cubic additionally has context of all the issues and pull requests, and can generate summaries with context of all of that. So it's very useful for peer/assigned reviewers to quickly catch up on what's happening.
Question related to: https://github.com/KTH/devops-course/pull/2700 Question: What do you think went wrong with cubic not being able to review the PR in time for the demo? Answer: This might be related to the LLM behind cubic taking longer time due to unknown constraints to review the PR.
Question related to: #2700 Breaking and Fixing a React App with CI Guards - Demo Q: Can Cubic identify bugs by itself, or does it only help fix bugs you have identified? A: It can actually find issues with the commits. But this can take a longer time. In big systems, it can identify many issues. It is not always that those issues are relevant. You can choose which of those issues you want to change.
Question on Demo #2705 What happens when some of the tasks fails, will the other independent tasks also fail?
Yes, we have to redo the whole pipeline again.
Presentation #2705 | Stop Rebuilding the Wheel: Accelerating Monorepos with Turborepo's Remote Cache
Q: You talk alot about monorepos, is there any reason we cannot, or should not use turborepo with non-monorepositories? Things like tests can still take a while to run even in non-monorepos.
A: While turborepo advertises itself as a solution for monorepos specifically, there is no inherent reason why you would not be able to use it in a non-monorepo project. Of course the benefits of using it would be greater if you have a monorepo. Turborepo themselves say that you can use it just fine with non-monorepos.
Question on presentation #2705
Q: it's considered a common practice to not store sensitive in code and instead to keep sensitive information in secret manager. So what other security concerns will be?
A: keep in mind that it's possible we might have sensitive information stored in the built artifacts, so it's better to think about the security aspect of it.
Question on #2705 Stop Rebuilding the Wheel: Accelerating Monorepos with Turborepo's Remote Cache
Q: Does Turborepo integrate with or replace npm? And does it work for languages other than JavaScript?
A: Turborepo is specific to JavaScript and Typescript and works in conjunction with npm, rather than replacing it.
Question on #2704 - Demo
Q: Is the pipline only configured on the main brnach
A: No, it's configured on all branches
Question to #2726 Demo title: Automated A/B Testing with Feature Flags
Q: You mentioned that half the users get one version and half get another. How does the split work? How does it determine who gets what? Is it random or can you specify who gets which version?
A: You can specify who gets which versions by geo location, time of day, etc... However, generally, for A/B testing, you want it to be as general as possible for most cases.
Question related to: Automated A/B Testing with Feature Flags #2726 - Demo Q: Is it necessary to use 50 % and 50 % split, or is it better to use some other percentages? A: No, you can use any other percentages, but it's good to have a large testing size.
question related to #2720 Q: is there some way to work around the block if needed due to special circumnstances A: you can force push using --no-verify flag
A question related to Blocking push by linting result and blocking commit by testing and todo-comment-checking results with flag #2720 .
Q: Do you think it could be a good idea to make the local hooks fix the errors (if possible) and not only complain about them? A: It could be something to make as an improvement, but there is high complexity in defining the error types to detect and fix.
Question on Fuzzing in your CI with OSS-Fuzz: detect unexepected bugs early. #2708
Q: How many times does it run? How long does it run for? A: Each time you commit, how often it runs depends on the configuration file, and it could potentially run 24 hours a day.
Question related to: https://github.com/KTH/devops-course/pull/2710 - Hyperfine Demo Q: Does it only measure time or are there other performance metrics? A: It only measures time, but there are options to make it more fine-grained, e.g. for multithreading.
Hyperfine performance test #2710
Q: How popular was this tool. Did you find many other places where it was used? A: It was not very popular but there were active members in the Github who were contributing. But Hyperfine wants to increase their users.
Story Design for CI Methodology (Presentation) - https://github.com/KTH/devops-course/pull/2707
Q: While it's good in principle, it might be tricky for teams to actually agrees on how long/short/complete the acceptance criteria should be? What's your take on this?
A: The idea of writing a more "complete" criteria is to spend more time now, so that more time could be saved later. So the criteria should be just "good enough" for the team to work on with more caution, although it doesn't mean that the story needs to be very complete.
Property-based testing: #2711
Q: Is property-based testing supposed to be replacement for unit tests, or a complement to it?
A: Depends. It is mainly supposed to be a complement.
canary releases #2740
Q: Is gradual rollout a defining feature of canary release?
A: It is configurable in AWS. It's not necessary but helps in the event of rollback if something breaks.
Question related to: https://github.com/KTH/devops-course/pull/2740
Question: What is the difference between Blue-Green-Deployment and Canary Releases
Answer: One of the main differences is that Blue-Green Deployment is more resource-intensive and more complex to set up. Next it is also possible to gradually rollout a version specifiying the region or prozent of traffic routed to a new release, which is not possible with Blue-Green-Deployment.
Presentation: Safe Friday Deployments (https://github.com/KTH/devops-course/pull/2740)
Q: Canary release reminds me of feature flags or A/B tests where a feature could only be accessed by a subset of users. In your opinion, in what situation should team consider the canary release vs feature flag approach?
A: Feature flag is more like a toggle where the team can turn on/off features -- so canary release might be more suitable if you have a "big update" to ensure everything work as a whole
Question related to: #2737
Question: Will the pipeline scan for newly added files? Answer: Yes, you can configure it to add this scans and adding new files also triggers the system to scan all the files during deployment.
continue deployment vs continuous delivery #2737
Q: If you could coin a different term so that they don't share the same acronym, what would it be?
A: Since both are established terms that millions of developers use, it would be hard to introduce a new term. Some people will say "CD deployment" or "CD delivery" to clarify.
Question related to #2737 (differences between continous deployment and continuous delivery)
Q: Which model do companies like Amazon, Netflix, or Google use more often?
A: They rely on deployment because frequent updates and speed is a priority and customers expect quick improvements. However, delivery is also used. For example if an intern makes a change and it passes all automated tests it might go to production. This may be risky since small mistakes can cause serious/big issues.
Presentation: GitOps: Streamlined CD through Git (#2734)
Q: does GitOps reduce the need of handling all sorts of Git Tokens when implementing DC operations?
A: yes using GitOps eliminates the need of using Token, compaired to when using Jenkins for example
Presentation: GitOps: Streamlined CD through Git (#2734)
Q: In the case that more manual review is required, isn't it doing a Pull Request will about to fulfill it?
A: its doable, but the the focus of this presentation is more related to GitOps.
Question to #2744
Demo title: Building and running CI/CD pipelines with TeamCity: A Live Demo
Q: Does TeamCity pipelines only build the pipeline outline or also run it? If it runs it, does this run locally?
A: You get the choice to do both. You can do on premise (can do more stuff for free). If useing cloud however (as done in demo), you don’t have as many build tokens. You can also decide on what it runs.