25-28. 4. 2017
Budapest
27 workshops
6 stages
1964 attendees
21 meetups
65 speakers
46 countries
Day 1, workshop:
Hypermedia hacking (Mike Amundsen - API Academy)
- Write clients as games - you can’t control how the user interacts with them, but you need to cover "all" the cases.
- Hypermedia is describing the actions/ providing instructions not in client code but in the messages.
- Action logic lives on backend - no need to update client code, when changing a feature.
- Non HTTP, domain specific actions (approve, reject, print ...) are problematic to implement with CRUD - we usually POST the action.
- SIREN (https://github.com/kevinswiber/siren)
Day 2, workshop:
Scaling technology and organisation together (Randy Shoup)
- Practices he experienced while working at Google, eBay etc.
-
3 phases of a company growth:
- search phase (prototype architecture, rapid iteration, as little technology as possible),
- execution phase ("just-enough", monolithic architecture, meet near-term, evolving customer needs as cheaply as possible),
- scaling phase (stay ahead of rapidly growing business, keep the site up - scale technology and teams).
Day 2, workshop:
Scaling technology and organisation together (Randy Shoup)
- Big successful companies are successful, because they work as many successful smaller teams.
- Team size might impact how system is designed (large team = monolith, smaller teams = "microservices").
-
15 to 50 people on the team is hard to divide. But as you grow, teams of 4-6 can start to own their services.
- full-stack teams
- end-to-end feature ownership
TEAM CULTURE is the most important factor for success!
(collaboration, learning, quality discipline etc. )
Day 3, talks:
Serverless Design Patterns with AWS Lambda: Big Data with Little Effort (Dr. Tim Wagner)
- Pitch for different Amazon services.
Decisions Decisions (Dan North)
- “Trade-offs are always happening and it is important to learn to navigate.”
- When you know what you are trading off you can make informed decisions.
Day 3, talks:
Tackling technical debt at scale (Yvette Pasqua - Meetup)
-
Designing a productive engineering culture:
- design for reliability,
- design for simplicity,
- design for speed (build times, deployments etc.).
-
At Meetup they created platform teams with focus to satisfy the goals for engineering culture:
- Not focused on product development.
- Focused on “continuous product health” (Test coverage, reduce build times, refactor code, better performance).
- Communicate the impact of cleaning/ refactoring work outside engineering team.
Day 3, talks:
Tackling technical debt at scale (Yvette Pasqua - Meetup)
- Low hanging fruits are not the best way to start the process. High impact tasks are much better (but harder).
-
4 high impact criteria:
-
hard cost (cloud, vendors)
-
platform reliability
-
system simplicity
-
development speed
-
Day 3, talks:
Scaling Lean: Principles over Process (Jeff Gothelf)
-
Principle #1: Customer value = Business value
- use OKRs
-
Principle #2: Value learning over delivery
- build momentum with pilot teams
- sandboxes and constraints for pilot team to experiment
- balance of learning/ delivery velocity
-
Principle #3: Radical transparency
- transparency through rituals (e.g. “stand-ups”, “scrum-of-scrums”, “demo days")
- access to customers
- access to data (analytics, usage data etc.)
Day 3, talks:
Scaling Lean: Principles over Process (Jeff Gothelf)
-
Principle #4: Humility in all things
- It is impossible to know exactly how a feature will look at the end - there’s no “done” in software development.
- Product discovery (don’t assume what to build, go discover it - with the least amount of work needed).
-
Small, agile teams that discover the product by themselves may not need managers, scrum masters etc.
-
Value prototyping.
Day 3, talks:
Building Great Teams: Culture and Core Protocols (Richard Kasperowski)
- Psychological safety - team members feel safe to take risks and be vulnerable in front of each other.
-
Teams protocol stack:
-
positive bias: non-negativity,
-
freedom (“I pass” when somebody doesn’t want to do something),
-
self-awareness about current emotional state,
-
connection -> check in or ask for help,
-
error handling.
-
Day 3, talks:
Building Great Teams: Culture and Core Protocols (Richard Kasperowski)
In high-performing teams people act as they are friends.
Day 3, talks:
Fun With Dead Languages (Damian Conway)
Day 4, talks:
The danger of team safety (Katherine Kirk)
-
3 characteristics of existence/ IT industry:
- everything is in a constant state of change,
- it's not about "the me",
- we'll always battle dissatisfaction - nothing is perfect.
-
It is hard to live "continuousness" - continuous delivery, continuous innovation.
-
Continuous delivery = continuous consequences, responsibility for deployed code
-
Continuous delivery = Continuous interaction/ collaboration
-
"Don't bother with safety. We are part of the industry that is revolutionary. Now is the time for persistence, resilience, bravery."
Day 4, talks:
Mutation Testing to the rescue of your tests (Nicolas Fränkel)
- Code coverage should not be the only metric for test quality.
-
Mutation testing:
- Mutate production code and run the same tests on it - tests should not pass.
- Example: tools (mutators) that automatically transform operators in your code.
Day 4, talks:
Machine Learning for Developers (Danilo Poccia)
- Machine Learning 101
-
Image generation with 2 neural networks:
- Learn by generating random images with one neural network and compare them to real images with another neural network.
- Over time generated images are getting more "real".
Day 4, talks:
Machine Learning for Developers (Danilo Poccia)
- Camera for the blind
Day 4, talks:
Better engineering via better discourse (Theo Schlossnagle)
"Exceptional communicators can impart great influence and value
... even to bad listeners."
"Exceptional listeners can extract value
... even from bad communicators."
"Bias in operation/ crisis can help you react quickly."
"Bias in learning is what makes you stupid."