Rich atoms don't cache suspended fields

[JaroslavTulach]

Suspended atom fields, as implemented by

• Suspended atom fields are evaluated only once #6151

may not be cached properly, when the amount of atom fields is higher than four. Following program shall cache the value generated by the java.util.Random generator, but it doesn't:

polyglot java import java.util.Random as R

type Num
    Holder a b c d ~num

    new  = Num.Holder "a" "b" "c" "d" (R.new.nextInt)

main =
   fbl = Num.new
   f = fbl.num
   n = fbl.num
   [ f, n, f == n ]

It is able to print [712182466, -522235373, False].

[JaroslavTulach]

Removing any of the a, b, c or d dummy atom fields fixes the problem: [-1299288440, -1299288440, True]

[JaroslavTulach]

Allocating the Num.Holder with primitive types like Num.Holder 1 2 3 4 (R.new.nextInt) doesn't fix the problem.

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2024-01-12):

Progress: - TCK PR integrated: #8685

• finishing rich atoms issue Rich atoms don't cache suspended fields #8710: b543c7c
• standups
• assert advice: Fix Runtime.assert #8742 (comment) It should be finished by 2024-01-16.

Next Day: Integrate Atoms, look towards BigDecimal support

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2024-01-14):

Progress: - Fixing atom NI failures: 6d890de

• merging to resolve conflicts:
• refactoring AtomNewInstanceNode & co.: 85c396f
• trying to write getUncached version
• AtomConstructorTest: d433218
• uncached variant: 8675747
• unboxlingLayouts optimized in uncached variant: f8d8a98 It should be finished by 2024-01-16.

Next Day: Integrate Atoms, look towards BigDecimal support

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2024-01-15):

Progress: - "Atom PR" is ready: #8712

• reviews test API: Refactor Table_Tests to the builder API #8622
• standups & demo meetings It should be finished by 2024-01-16.

Next Day: Review Test API, look towards BigDecimal support

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2024-01-16):

Progress: - "Atom PR" integrated: #8712

• deep review of Test API: Refactor Table_Tests to the builder API #8622 (comment)
• review with Pavel. Test API & Atom constructors.
• derive --in-project: Derive --in-project from --run source location #8775
• working on Complex number specs: Binary operator resolution based on that value #8779 It should be finished by 2024-01-16.

Next Day: BigDecimal (via Complex number) support