This seems like a very similar issue to the various issues about the `=` in
keyword argument parsing to a different expression than `=` in an
assignment. They print equal, but parse to different expressions, depending
on the context. Not sure what Jeff want to do, but it seems likely that we
ultimately would like to use the same expressions for the same syntactic
elements.
Ivar
kl. 12:54:43 UTC+2 tirsdag 21. oktober 2014 skrev Mauro følgende:
>
> > What is Expr(:<:, :X, :Int) supposed to mean?
>
> It's part of a function definition:
>
> :(f{X<:Int}(x::X) = 5).args[1].args[1].args[2]
>
> I came across this when doing unit testing for a function which picks a
> function definition apart:
>
> julia> parsef(fn) = fn.args[1].args[1].args[2]
> parsef (generic function with 1 method)
>
> julia> fn = :(f{X<:Int}(x::X) = 5)
> :((f{X<:Int})(x::X) = 5)
>
> julia> parsef(fn)
> :(X<:Int)
>
> julia> @test parsef(fn)==:(X<:Int)
> ERROR: test failed: (X<:Int == X <: Int)
>
> julia> @test parsef(fn)==Expr(:<:, :X, :Int)
> # passes test
>
> >
> > I just get an AST error when trying to evaluate that expression, so it
> > can't be very usefull
> >
> > julia> eval(Expr(:<:, :Int64, :Int))
> > ERROR: unsupported or misplaced expression <:
> >
> > We should probably consider erroring out when printing invalid
> expressions.
> >
> > Regards
> > Ivar
> >
> > kl. 11:05:34 UTC+2 tirsdag 21. oktober 2014 skrev Mauro følgende:
> >>
> >> This seems a bit confusing:
> >>
> >> ```
> >> julia> Expr(:<:, :X, :Int)
> >> :(X<:Int)
> >>
> >> julia> Expr(:<:, :X, :Int).head
> >> :<:
> >>
> >> julia> :(X<:Int)
> >> :(X <: Int)
> >>
> >> julia> :(X<:Int).head
> >> :comparison
> >> ```
> >>
> >> Is this intentional? This also means that copy-pasting the string
> >> representation of `Expr(:<:, :X, :Int)` does not yield itself.
> >>
>
>
