[meeting] [joint] 14.06.2024

[deemp]

Almost a transcript

• Nickolay:
  • Last meeting Yegor told they'll fix opeo on their side
  • Don't ping them
  • We can help!
• Vladimir:
  • Agree, don't need to ping
• Nickolay:
  • Need a good example with wrappers (Compose an EO program that can be normalized via user-defined rules #404)
  • Applied, but no optimizations (Try normalization with user-defined rules only #415)
  • Yegor should show how multiplication table should work
  • We don't dataize stdout
  • Yegor told we don't want to dataize inside atoms. Need a general mechanism that doesn't know about an atom
  • We don't know what parts of rho are necessary
  • It's a conceptual problem
• Yegor:
  • We can improve when there are a lot of abstract classes
  • We need a program with three abstract (e.g.) objects with several arguments
• Nickolay:
  • We need a program in Java or EO
  • It may be a blocker
  • Violetta sent a PR
    • Use context
    • Didn't explain what's a context
• Yegor:
  • Need to make soft for theory
• Nickolay:
  • We're guided by concrete examples and rules that we want
• Yegor:
  • CI fails
  • We need calculus to be minimal
  • Rewriting rules should be minimal
  • New rules are unreadable
• Nickolay:
  • Violetta wants to provide details
  • Yegor wants readability
  • Lambda calculus also has context
• Yegor:
  • Violetta should make the PR to build
• Nickolay:
  • Bound variables in outer scope
  • If typed, type information
  • In PHI, we store the object where we are
• Yegor:
  • Rules were readable
• Nickolay:
  • Can write just about the information in the context to make sense of rules
• Violetta:
  • Some rules depend on what's the current object
• Nickolay:
  • Can use conditions
• Yegor:
  • Will share rules, Violetta should explain what's wrong

• Violetta:
  • How to work with rules where the context changes?
• Yegor:
  • Show the simplest rule
• Violetta:
  • Rule: if global object dot something, then this case should be processed
• Nickolay
  • Rule: Phi.attribute - where's this rule?
  • R_DOT: object.tau and A(C ...)
  • Where do you unpack Phi.something?
  • You want to Something.attribute to go there, unwrap and go to bytes
  • In normalization - we want to extract an attribute
• Nickolay:
  • Violetta should add examples.
• Yegor:
  • Changes are too big.
  • Can't build the PR..
• Nickolay:
  • We do use contextualization and adoption.
  • We use a different notation.
• Violetta:
  • Operations are used simultaneously, that's why they're united.
• Nickolay:
  • Yegor has rules without adoption.
  • In our contextualization, we don't do precopy.
  • We have a condition that an object inserted into another object is in the normal form.
  • If an object is in the normal form, contextualization won't change anything.
  • Try to not change notation.
  • Provide examples.
  • Expects to agree on rules till the end of June
• Volodya:
  • Will send examples
• Nickolay:
  • Yegor should send an example
  • Can meet if necessary
  • Want to use global object
  • Any expectations from the normalizer?
  • We have proposals for user-defined rules.
  • If don't need to use the normalizer with spring, please tell
• Yegor:
  • Spring - will be ready in several weeks
  • Will ask Max to send a concrete rule for ineo.
• Nickolay
  • Provide a concrete example for user-defined rules