1.7.2 errors with Agda 2.6.4

Question

1.7.2 errors with Agda 2.6.4

juhp opened this issue 9 months ago · comments

This may be my fault, but I tried to rebuild the Fedora agda-stdlib 1.7.2 package with Agda 2.6.4 and it seems to fail:

Reason: It relies on injectivity of the data constructor +_, which
        is not yet supported

when checking the definition of neg-antimono-≤
/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Properties.agda:726,1-732,97
warning: -W[no]UnsupportedIndexedMatch
This clause uses pattern-matching features that are not yet
supported by Cubical Agda, the function to which it belongs will
not compute when applied to transports.

Reason: It relies on injectivity of the data constructor -[1+_],
        which is not yet supported

when checking the definition of neg-antimono-≤
/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Properties.agda:805,1-824,80
warning: -W[no]UnsupportedIndexedMatch
This clause uses pattern-matching features that are not yet
supported by Cubical Agda, the function to which it belongs will
not compute when applied to transports.

Reason: It relies on injectivity of the data constructor false,
        which is not yet supported

when checking the definition of Data.Rational.Properties.with-2764
/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Properties.agda:805,1-824,80
warning: -W[no]UnsupportedIndexedMatch
This clause uses pattern-matching features that are not yet
supported by Cubical Agda, the function to which it belongs will
not compute when applied to transports.

Reason: It relies on injectivity of the data constructor true,
        which is not yet supported

when checking the definition of Data.Rational.Properties.with-2764
/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Properties.agda:1029,1-1047,80
warning: -W[no]UnsupportedIndexedMatch
This clause uses pattern-matching features that are not yet
supported by Cubical Agda, the function to which it belongs will
not compute when applied to transports.

Reason: It relies on injectivity of the data constructor false,
        which is not yet supported

when checking the definition of Data.Rational.Properties.with-3110
/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Properties.agda:1029,1-1047,80
warning: -W[no]UnsupportedIndexedMatch
This clause uses pattern-matching features that are not yet
supported by Cubical Agda, the function to which it belongs will
not compute when applied to transports.

Reason: It relies on injectivity of the data constructor true,
        which is not yet supported

when checking the definition of Data.Rational.Properties.with-3110
 Checking Data.Rational.Instances (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Instances.agda).
 Checking Data.Rational.Literals (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Literals.agda).
 Checking Data.Rational.Show (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Show.agda).
 Checking Data.Rational.Solver (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Solver.agda).
 Checking Data.Rational.Unnormalised (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Unnormalised.agda).
 Checking Data.Rational.Unnormalised.Solver (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Rational/Unnormalised/Solver.agda).
 Checking Data.Record (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Record.agda).
 Checking Data.Refinement (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Refinement.agda).
 Checking Data.Refinement.Relation.Unary.All (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Refinement/Relation/Unary/All.agda).
 Checking Data.Sign.Instances (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Sign/Instances.agda).
 Checking Data.Star.BoundedVec (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Star/BoundedVec.agda).
  Checking Data.Star.Nat (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Star/Nat.agda).
   Checking Relation.Binary.Construct.Always (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Relation/Binary/Construct/Always.agda).
    Checking Relation.Binary.Construct.Constant (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Relation/Binary/Construct/Constant.agda).
/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Star/Nat.agda:47,1-49,26
warning: -W[no]UnsupportedIndexedMatch
This clause uses pattern-matching features that are not yet
supported by Cubical Agda, the function to which it belongs will
not compute when applied to transports.

Reason: It relies on injectivity of the data constructor tt, which
        is not yet supported

when checking the definition of _∸_
/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Star/Nat.agda:47,1-49,26
warning: -W[no]UnsupportedIndexedMatch
This clause uses pattern-matching features that are not yet
supported by Cubical Agda, the function to which it belongs will
not compute when applied to transports.

Reason: It relies on injectivity of the data constructor tt, which
        is not yet supported

when checking the definition of _∸_
  Checking Data.Star.Decoration (/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Star/Decoration.agda).
/var/home/petersen/fedora/haskell/Agda-stdlib/BUILD/agda-stdlib-1.7.2/src/Data/Star/Decoration.agda:32,3-4
Set ℓ is not less or equal than Set p
when checking that the type I of an argument to the constructor ↦
fits in the sort Set p of the datatype.
Note: this argument is forced by the indices of ↦, so this
definition would be allowed under --large-indices.

Maybe only the last one is an actual error?

MatthewDaggitt · Answer 1 · Tue Oct 10 2023 08:01:28 GMT+0800 (China Standard Time)

Calling @plt-amy . Is there anything that has changed in cubical with indices in the last release?

shuhung · Answer 2 · Tue Oct 10 2023 10:31:10 GMT+0800 (China Standard Time)

@plt-amy fixed large index issues of stdlib in #2030 to accommodate Agda 2.6.4 changes

Jens Petersen · Answer 3 · Tue Oct 10 2023 12:46:44 GMT+0800 (China Standard Time)

So a new release is needed?

Andreas Abel · Answer 4 · Tue Oct 10 2023 16:15:42 GMT+0800 (China Standard Time)

Ah, indeed, I didn't test all of 1.7.1/2 with Agda 2.6.4, the parts of it that my projects used checked fine with Agda 2.6.4.
But indeed, some fixing is needed (and then a release of 1.7.3).
The minimal patch for 1.7.2 is to add to option --large-indices to the flags section of the .agda-lib file, but this is of course some overkill, as only some files need it.
The better way is to backport the fix that @plt-amy applied to the master of the std-lib, pointed out by @shhyou:

#2030

jamesmckinna · Answer 5 · Tue Oct 10 2023 16:37:42 GMT+0800 (China Standard Time)

Well, if we are talking about back-porting/back-patching v1.7.2, then I would advocate strongly for a revised version of #2125 for v.1.7.2 (or this hypothetical v1.7.3 which you allude to here) to supply enough Effect.* stubs to resolve issue #2124 ...

Andreas Abel · Answer 6 · Tue Oct 10 2023 16:44:22 GMT+0800 (China Standard Time)

@jamesmckinna: Good idea.

I'll give v1.7.3 a shot...

MatthewDaggitt · Answer 7 · Wed Oct 11 2023 09:54:38 GMT+0800 (China Standard Time)

Thanks @andreasabel !

MatthewDaggitt · Answer 8 · Fri Oct 13 2023 09:27:41 GMT+0800 (China Standard Time)

@juhp this is fixed in the just released v1.7.3. Apologies that this slipped under our radar.