Polar talks: code generation
Download as PPTX, PDF0 likes49 views
The document discusses code generation and meta programming, exploring their definitions, advantages, and challenges. It highlights the productivity gains and simplifications offered by these practices, while also addressing concerns about complexity and maintenance issues. The discussion emphasizes the evolution of code generation techniques and poses questions about the future of software development and the potential for self-modifying applications.
Polar talks: code generation
- 1. Code Generation for the New Generation
- 2. Introductions Tome Gramatikov Singular Dejan Jankov Seavus Risto Kavalinovski Polar Cape
- 3. Motivation for this discussion ● What is code generation? And what is Meta programming? ● Good or Bad ● Past present and future aspects Twitter questions: #PolarTalks
- 4. We all did it.. ● Do you know you use it? ● Frameworks, Libraries ● Dynamic languages ○ Ruby, Python, etc. ● SOAP / REST / JPA / AOP ○ GraphQL, Hibernate, Swagger, Apache Thrift ● BPM ● IDEs to bootstrap applications
- 5. Benefits ● We gain: ○ Productivity ○ Simplification ○ Portability ○ Consistency ● BoilerHate :) ● Software manufacturing ● Clone yourself
- 6. Code generation sucks? ● Introduces ○ Complex not optimized code ○ Difficulty during maintenance ● Problems we had in the past, ○ why is it not good? ● Has code gen helped you or not? ● Is it worth to generate one time job?
- 7. How dynamic is Java ● late binding / method overrides ● Class loading at runtime ● Reflection API ● invokedynamic
- 8. Future of code generation ● Model driven design .. or business driven? ○ Goal driven? ● Hardware is becoming faster cheaper, ○ how about our languages/ coding? ● (When) Are we finally going to create apps without coding? ● Should the software modify / build itself?
- 9. Final Thoughts for the New Generation
Editor's Notes
- #4: A Guide to Code Generation - https://tomassetti.me/code-generation/ Code vs Data (Metaprogramming) - Computerphile - https://www.youtube.com/watch?v=dw-y3vNDRWk Raymond Hettinger - Dataclasses: The code generator to end all code generators (first few minutes only) - https://www.youtube.com/watch?v=T-TwcmT6Rcw A Practical Guide to Metaprogramming - https://www.youtube.com/watch?v=lZfv4H-9ato
- #5: Tip of the iceberg: what it looks like when we use tools and frameworks, but what’s under the hood?
- #6: How close are we to start manufacturing the apps, repro materials on input, use recipe and deliver a product as output? Can we clone part of ourselves that write code using the code generators, automate repetitive tasks?
- #9: Model driven design only generates the classes and relationships, not behaviors which we actually need? It is sometimes forgotten that the framework that allows elimination of traditional programming (code-gen) in an area is a system or library that has been designed, programmed, and tested in the traditional way.