Re: [julia-users] getindex for a real number
> > Sure, but it should be one way or the other, not half way between. > +100 On Wednesday, December 2, 2015 at 6:51:35 PM UTC+1, Cedric St-Jean wrote: > > There's also > > x = [1 2; 3 4] > x[:,[1]] # returns 2D array > x[:, 1] # returns 1D array > x[1, :] # AFAIK, returns 2D under Julia 0.4 and 1D under 0.5 > > I like the 0.5 behaviour a lot better. > > On Wednesday, December 2, 2015 at 12:21:34 PM UTC-5, Tim Holy wrote: >> >> Glen, that's a great list of bugs. Have you considered filing them as >> issue(s)? >> >> Some immediate thoughts: >> >> On Wednesday, December 02, 2015 06:17:06 AM Glen O wrote: >> > As an example, reshape(1,1) throws an error >> >> I'm not sure that's a real problem, although indeed implementing reshape >> on >> numbers would be more efficient than reshape([1], (1,1,1)) because in the >> latter >> you're creating two arrays. So possibly this is something we should >> implement. >> >> > , and squeeze(1,(1,)) gets stuck >> > in an infinite loop. >> >> That's definitely a bug. It's surely a very slow stack overflow (infinite >> recursion). >> >> > vec(1) >> >> Similar to reshape...maybe/maybe not. >> >> > throws an error, as does cumsum(1). >> >> Since sum(1) works, this should too. Bug. >> >> > And of >> > course there's the issue with getindex involving colon, arrays or >> ranges >> > for indexing (you'd think that, just as a[[1,1]] gives the value of >> a[1] >> > twice for an array, that it would do the same for a scalar, but it >> doesn't). >> >> Bug >> >> > I can understand the desire not to have them be identical (since there >> are >> > cases where a function should do a different thing for a number than it >> > does for an array), yet allow partial compatibility... it's just a >> little >> > arbitrary which cases work and which don't. >> >> Reports would help---not everyone hits these (I'm not sure I ever have). >> >>
Re: [julia-users] getindex for a real number
There's also x = [1 2; 3 4] x[:,[1]] # returns 2D array x[:, 1] # returns 1D array x[1, :] # AFAIK, returns 2D under Julia 0.4 and 1D under 0.5 I like the 0.5 behaviour a lot better. On Wednesday, December 2, 2015 at 12:21:34 PM UTC-5, Tim Holy wrote: > > Glen, that's a great list of bugs. Have you considered filing them as > issue(s)? > > Some immediate thoughts: > > On Wednesday, December 02, 2015 06:17:06 AM Glen O wrote: > > As an example, reshape(1,1) throws an error > > I'm not sure that's a real problem, although indeed implementing reshape > on > numbers would be more efficient than reshape([1], (1,1,1)) because in the > latter > you're creating two arrays. So possibly this is something we should > implement. > > > , and squeeze(1,(1,)) gets stuck > > in an infinite loop. > > That's definitely a bug. It's surely a very slow stack overflow (infinite > recursion). > > > vec(1) > > Similar to reshape...maybe/maybe not. > > > throws an error, as does cumsum(1). > > Since sum(1) works, this should too. Bug. > > > And of > > course there's the issue with getindex involving colon, arrays or ranges > > for indexing (you'd think that, just as a[[1,1]] gives the value of a[1] > > twice for an array, that it would do the same for a scalar, but it > doesn't). > > Bug > > > I can understand the desire not to have them be identical (since there > are > > cases where a function should do a different thing for a number than it > > does for an array), yet allow partial compatibility... it's just a > little > > arbitrary which cases work and which don't. > > Reports would help---not everyone hits these (I'm not sure I ever have). > >
Re: [julia-users] getindex for a real number
Glen, that's a great list of bugs. Have you considered filing them as issue(s)? Some immediate thoughts: On Wednesday, December 02, 2015 06:17:06 AM Glen O wrote: > As an example, reshape(1,1) throws an error I'm not sure that's a real problem, although indeed implementing reshape on numbers would be more efficient than reshape([1], (1,1,1)) because in the latter you're creating two arrays. So possibly this is something we should implement. > , and squeeze(1,(1,)) gets stuck > in an infinite loop. That's definitely a bug. It's surely a very slow stack overflow (infinite recursion). > vec(1) Similar to reshape...maybe/maybe not. > throws an error, as does cumsum(1). Since sum(1) works, this should too. Bug. > And of > course there's the issue with getindex involving colon, arrays or ranges > for indexing (you'd think that, just as a[[1,1]] gives the value of a[1] > twice for an array, that it would do the same for a scalar, but it doesn't). Bug > I can understand the desire not to have them be identical (since there are > cases where a function should do a different thing for a number than it > does for an array), yet allow partial compatibility... it's just a little > arbitrary which cases work and which don't. Reports would help---not everyone hits these (I'm not sure I ever have).
Re: [julia-users] getindex for a real number
On Wednesday, December 02, 2015 05:11:32 AM Seth wrote:
> This is great. I assume it catches "for i in n" where n is a scalar, also,
> right?
>
> I could see requiring this by default in all my packages.
Might be better as a test dependency, since it works by "breaking" methods in
Base. If you include it in your package by default, you risk fragmenting Julia
into incompatible dialects.
--Tim
>
> On Wednesday, December 2, 2015 at 1:49:22 AM UTC-8, Eric Forgy wrote:
> > Some more improvements...
> >
> > julia> n = 5
> > 5
> >
> > julia> for i = n
> >
> >println(i)
> >end
> >
> > 5
> >
> > julia> using strict
> >
> > julia> for i = n
> >
> >println(i)
> >end
> >
> > ERROR: MethodError: `start` has no method matching start(::Type{Number})
> >
> > in start at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:24
> >
> > On Wednesday, December 2, 2015 at 12:05:07 PM UTC+8, Eric Forgy wrote:
> >> It's a start :)
> >>
> >> https://github.com/EricForgy/strict.jl
> >>
> >>
> >> julia> using strict
> >>
> >> julia> a = 5
> >> 5
> >>
> >> julia> a[1]
> >> ERROR: MethodError: `getindex` has no method matching getindex(::Type{
> >> Number}, ::Type{Integer})
> >>
> >> Closest candidates are:
> >> getindex(::Type{T}, ::Any...)
> >> getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{
> >>
> >> T})
> >>
> >> getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{
> >>
> >> T}, ::Range{T}...)
> >>
> >> in getindex at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:4
> >>
> >> On Wednesday, December 2, 2015 at 11:02:45 AM UTC+8, Tim Holy wrote:
> >>> Likewise, I do see why this is a little troublesome. It's annoying when
> >>> you
> >>> mean to write `for i = 1:n` but accidentally write `for i = n`; it's not
> >>> always an easy bug to find.
> >>>
> >>> --Tim
> >>>
> >>> On Tuesday, December 01, 2015 06:38:46 PM Eric Forgy wrote:
> >>> > It bugs me, but only a little, so I won't lose sleep over it :)
> >>> >
> >>> > Then again, I wish Julia had a "strict" mode. In strict mode, the
> >>>
> >>> language
> >>>
> >>> > would be more pure mathematically, e.g. scalars have no indices, the
> >>> > transpose of a vector is a covector, etc. This bit me recently because
> >>>
> >>> if T
> >>>
> >>> > <: U, then Array{T} is NOT <: Array{U} although as, sub-modules
> >>>
> >>> > , Tmodule <:
> >>> Umodule.
> >>>
> >>> > Then again, as I'm learning, if we want Julia to do something bad
> >>>
> >>> enough,
> >>>
> >>> > e.g. have a "strict" mode, we can have it. For example, I could write
> >>>
> >>> a
> >>>
> >>> > package "strict.jl" where
> >>> >
> >>> > using strict
> >>> >
> >>> > would kill Base.getindex(::Number) and things like that. That could be
> >>>
> >>> cool
> >>>
> >>> > :)
> >>> >
> >>> > On Wednesday, December 2, 2015 at 9:38:50 AM UTC+8, Tim Holy wrote:
> >>> > > On Tuesday, December 01, 2015 03:19:33 PM Eric Forgy wrote:
> >>> > > > A scalar is distinct from a vector so size(a) = () makes sense.
> >>>
> >>> getindex
> >>>
> >>> > > for
> >>> > >
> >>> > > > a scalar does not make sense and should probably be removed on the
> >>> > >
> >>> > > grounds
> >>> > >
> >>> > > > of mathematical elegance :) Any code that depends on referencing a
> >>> > >
> >>> > > scalar
> >>> > >
> >>> > > > via an index is probably flawed in the first place.
> >>> > >
> >>> > > Conversely, there are many people who seem to want Julia to treat
> >>>
> >>> scalars
> >>>
> >>> > > and
> >>> > > 1-vectors indistinguishably (ala Matlab).
> >>> > >
> >>> > > For what it's worth, here's a (contrived) example to justify the
> >>>
> >>> current
> >>>
> >>> > > behavior:
> >>> > >
> >>> > > function sum_over_dims(A, dims)
> >>> > >
> >>> > > for d in dims
> >>> > >
> >>> > > A = sum(A, d)
> >>> > >
> >>> > > end
> >>> > > A
> >>> > >
> >>> > > end
> >>> > >
> >>> > > sum_over_dims(A, [2,3])
> >>> > > sum_over_dims(A, 2)
> >>> > >
> >>> > > Why should I write sum_over_dims(A, [2]) in the latter case?
> >>> > >
> >>> > > Best,
> >>> > > --Tim
Re: [julia-users] getindex for a real number
On Wednesday, 2 December 2015 11:38:50 UTC+10, Tim Holy wrote: > > Conversely, there are many people who seem to want Julia to treat scalars > and > 1-vectors indistinguishably (ala Matlab). > Sure, but it should be one way or the other, not half way between. In Matlab, scalars are literally 1-vectors, so everything that works with 1-vectors works with scalars. In Julia, some things work, and others don't. That just results in confusion. As an example, reshape(1,1) throws an error, and squeeze(1,(1,)) gets stuck in an infinite loop. vec(1) throws an error, as does cumsum(1). And of course there's the issue with getindex involving colon, arrays or ranges for indexing (you'd think that, just as a[[1,1]] gives the value of a[1] twice for an array, that it would do the same for a scalar, but it doesn't). I can understand the desire not to have them be identical (since there are cases where a function should do a different thing for a number than it does for an array), yet allow partial compatibility... it's just a little arbitrary which cases work and which don't.
Re: [julia-users] getindex for a real number
On Tuesday, December 1, 2015 at 9:38:46 PM UTC-5, Eric Forgy wrote: > > Then again, I wish Julia had a "strict" mode. In strict mode, the language > would be more pure mathematically, e.g. scalars have no indices, the > transpose of a vector is a covector, etc. > Not all vectors represent elements of a Hilbert space, and not all 2d arrays represent linear operators. Sometimes arrays are just containers, for which those algebraic operations aren't sensible. There's nothing "impure" mathematically about this, any more than there is something "impure" about a broadcasting operation like .* ... mathematics encompasses more than linear algebra. The reason that scalars are iterable is that it can make it easier to write generic code, where the same code works if the argument is a single number or a container of numbers. The question is not whether this is a valid operation in linear algebra, but rather whether the convenience of being able to treat numbers as indexable length-1 containers outweighs the potential for confusion. Reasonable people can disagree about this, but let's please not get into metaphysical debates about purity.
Re: [julia-users] getindex for a real number
This is great. I assume it catches "for i in n" where n is a scalar, also,
right?
I could see requiring this by default in all my packages.
On Wednesday, December 2, 2015 at 1:49:22 AM UTC-8, Eric Forgy wrote:
>
> Some more improvements...
>
> julia> n = 5
> 5
>
> julia> for i = n
>println(i)
>end
> 5
>
> julia> using strict
>
> julia> for i = n
>println(i)
>end
> ERROR: MethodError: `start` has no method matching start(::Type{Number})
> in start at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:24
>
>
> On Wednesday, December 2, 2015 at 12:05:07 PM UTC+8, Eric Forgy wrote:
>>
>> It's a start :)
>>
>> https://github.com/EricForgy/strict.jl
>>
>>
>> julia> using strict
>>
>> julia> a = 5
>> 5
>>
>> julia> a[1]
>> ERROR: MethodError: `getindex` has no method matching getindex(::Type{
>> Number}, ::Type{Integer})
>> Closest candidates are:
>> getindex(::Type{T}, ::Any...)
>> getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{
>> T})
>> getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{
>> T}, ::Range{T}...)
>> in getindex at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:4
>>
>>
>> On Wednesday, December 2, 2015 at 11:02:45 AM UTC+8, Tim Holy wrote:
>>>
>>> Likewise, I do see why this is a little troublesome. It's annoying when
>>> you
>>> mean to write `for i = 1:n` but accidentally write `for i = n`; it's not
>>> always an easy bug to find.
>>>
>>> --Tim
>>>
>>> On Tuesday, December 01, 2015 06:38:46 PM Eric Forgy wrote:
>>> > It bugs me, but only a little, so I won't lose sleep over it :)
>>> >
>>> > Then again, I wish Julia had a "strict" mode. In strict mode, the
>>> language
>>> > would be more pure mathematically, e.g. scalars have no indices, the
>>> > transpose of a vector is a covector, etc. This bit me recently because
>>> if T
>>> > <: U, then Array{T} is NOT <: Array{U} although as, sub-modules
>>> > , Tmodule <:
>>> Umodule.
>>> >
>>> > Then again, as I'm learning, if we want Julia to do something bad
>>> enough,
>>> > e.g. have a "strict" mode, we can have it. For example, I could write
>>> a
>>> > package "strict.jl" where
>>> >
>>> > using strict
>>> >
>>> > would kill Base.getindex(::Number) and things like that. That could be
>>> cool
>>> >
>>> > :)
>>> >
>>> > On Wednesday, December 2, 2015 at 9:38:50 AM UTC+8, Tim Holy wrote:
>>> > > On Tuesday, December 01, 2015 03:19:33 PM Eric Forgy wrote:
>>> > > > A scalar is distinct from a vector so size(a) = () makes sense.
>>> getindex
>>> > >
>>> > > for
>>> > >
>>> > > > a scalar does not make sense and should probably be removed on the
>>> > >
>>> > > grounds
>>> > >
>>> > > > of mathematical elegance :) Any code that depends on referencing a
>>> > >
>>> > > scalar
>>> > >
>>> > > > via an index is probably flawed in the first place.
>>> > >
>>> > > Conversely, there are many people who seem to want Julia to treat
>>> scalars
>>> > > and
>>> > > 1-vectors indistinguishably (ala Matlab).
>>> > >
>>> > > For what it's worth, here's a (contrived) example to justify the
>>> current
>>> > > behavior:
>>> > >
>>> > > function sum_over_dims(A, dims)
>>> > >
>>> > > for d in dims
>>> > >
>>> > > A = sum(A, d)
>>> > >
>>> > > end
>>> > > A
>>> > >
>>> > > end
>>> > >
>>> > > sum_over_dims(A, [2,3])
>>> > > sum_over_dims(A, 2)
>>> > >
>>> > > Why should I write sum_over_dims(A, [2]) in the latter case?
>>> > >
>>> > > Best,
>>> > > --Tim
>>>
>>>
Re: [julia-users] getindex for a real number
Seems like this has the makings of a nice package for debugging.
Best,
--Tim
On Wednesday, December 02, 2015 01:49:22 AM Eric Forgy wrote:
> Some more improvements...
>
> julia> n = 5
> 5
>
> julia> for i = n
>println(i)
>end
> 5
>
> julia> using strict
>
> julia> for i = n
>println(i)
>end
> ERROR: MethodError: `start` has no method matching start(::Type{Number})
> in start at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:24
>
> On Wednesday, December 2, 2015 at 12:05:07 PM UTC+8, Eric Forgy wrote:
> > It's a start :)
> >
> > https://github.com/EricForgy/strict.jl
> >
> >
> > julia> using strict
> >
> > julia> a = 5
> > 5
> >
> > julia> a[1]
> > ERROR: MethodError: `getindex` has no method matching getindex(::Type{
> > Number}, ::Type{Integer})
> >
> > Closest candidates are:
> > getindex(::Type{T}, ::Any...)
> > getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{T
> >
> > })
> >
> > getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{T
> >
> > }, ::Range{T}...)
> >
> > in getindex at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:4
> >
> > On Wednesday, December 2, 2015 at 11:02:45 AM UTC+8, Tim Holy wrote:
> >> Likewise, I do see why this is a little troublesome. It's annoying when
> >> you
> >> mean to write `for i = 1:n` but accidentally write `for i = n`; it's not
> >> always an easy bug to find.
> >>
> >> --Tim
> >>
> >> On Tuesday, December 01, 2015 06:38:46 PM Eric Forgy wrote:
> >> > It bugs me, but only a little, so I won't lose sleep over it :)
> >> >
> >> > Then again, I wish Julia had a "strict" mode. In strict mode, the
> >>
> >> language
> >>
> >> > would be more pure mathematically, e.g. scalars have no indices, the
> >> > transpose of a vector is a covector, etc. This bit me recently because
> >>
> >> if T
> >>
> >> > <: U, then Array{T} is NOT <: Array{U} although as, sub-modules
> >>
> >> > , Tmodule <:
> >> Umodule.
> >>
> >> > Then again, as I'm learning, if we want Julia to do something bad
> >>
> >> enough,
> >>
> >> > e.g. have a "strict" mode, we can have it. For example, I could write
> >>
> >> a
> >>
> >> > package "strict.jl" where
> >> >
> >> > using strict
> >> >
> >> > would kill Base.getindex(::Number) and things like that. That could be
> >>
> >> cool
> >>
> >> > :)
> >> >
> >> > On Wednesday, December 2, 2015 at 9:38:50 AM UTC+8, Tim Holy wrote:
> >> > > On Tuesday, December 01, 2015 03:19:33 PM Eric Forgy wrote:
> >> > > > A scalar is distinct from a vector so size(a) = () makes sense.
> >>
> >> getindex
> >>
> >> > > for
> >> > >
> >> > > > a scalar does not make sense and should probably be removed on the
> >> > >
> >> > > grounds
> >> > >
> >> > > > of mathematical elegance :) Any code that depends on referencing a
> >> > >
> >> > > scalar
> >> > >
> >> > > > via an index is probably flawed in the first place.
> >> > >
> >> > > Conversely, there are many people who seem to want Julia to treat
> >>
> >> scalars
> >>
> >> > > and
> >> > > 1-vectors indistinguishably (ala Matlab).
> >> > >
> >> > > For what it's worth, here's a (contrived) example to justify the
> >>
> >> current
> >>
> >> > > behavior:
> >> > >
> >> > > function sum_over_dims(A, dims)
> >> > >
> >> > > for d in dims
> >> > >
> >> > > A = sum(A, d)
> >> > >
> >> > > end
> >> > > A
> >> > >
> >> > > end
> >> > >
> >> > > sum_over_dims(A, [2,3])
> >> > > sum_over_dims(A, 2)
> >> > >
> >> > > Why should I write sum_over_dims(A, [2]) in the latter case?
> >> > >
> >> > > Best,
> >> > > --Tim
Re: [julia-users] getindex for a real number
Some more improvements...
julia> n = 5
5
julia> for i = n
println(i)
end
5
julia> using strict
julia> for i = n
println(i)
end
ERROR: MethodError: `start` has no method matching start(::Type{Number})
in start at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:24
On Wednesday, December 2, 2015 at 12:05:07 PM UTC+8, Eric Forgy wrote:
>
> It's a start :)
>
> https://github.com/EricForgy/strict.jl
>
>
> julia> using strict
>
> julia> a = 5
> 5
>
> julia> a[1]
> ERROR: MethodError: `getindex` has no method matching getindex(::Type{
> Number}, ::Type{Integer})
> Closest candidates are:
> getindex(::Type{T}, ::Any...)
> getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{T
> })
> getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{T
> }, ::Range{T}...)
> in getindex at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:4
>
>
> On Wednesday, December 2, 2015 at 11:02:45 AM UTC+8, Tim Holy wrote:
>>
>> Likewise, I do see why this is a little troublesome. It's annoying when
>> you
>> mean to write `for i = 1:n` but accidentally write `for i = n`; it's not
>> always an easy bug to find.
>>
>> --Tim
>>
>> On Tuesday, December 01, 2015 06:38:46 PM Eric Forgy wrote:
>> > It bugs me, but only a little, so I won't lose sleep over it :)
>> >
>> > Then again, I wish Julia had a "strict" mode. In strict mode, the
>> language
>> > would be more pure mathematically, e.g. scalars have no indices, the
>> > transpose of a vector is a covector, etc. This bit me recently because
>> if T
>> > <: U, then Array{T} is NOT <: Array{U} although as, sub-modules
>> > , Tmodule <:
>> Umodule.
>> >
>> > Then again, as I'm learning, if we want Julia to do something bad
>> enough,
>> > e.g. have a "strict" mode, we can have it. For example, I could write
>> a
>> > package "strict.jl" where
>> >
>> > using strict
>> >
>> > would kill Base.getindex(::Number) and things like that. That could be
>> cool
>> >
>> > :)
>> >
>> > On Wednesday, December 2, 2015 at 9:38:50 AM UTC+8, Tim Holy wrote:
>> > > On Tuesday, December 01, 2015 03:19:33 PM Eric Forgy wrote:
>> > > > A scalar is distinct from a vector so size(a) = () makes sense.
>> getindex
>> > >
>> > > for
>> > >
>> > > > a scalar does not make sense and should probably be removed on the
>> > >
>> > > grounds
>> > >
>> > > > of mathematical elegance :) Any code that depends on referencing a
>> > >
>> > > scalar
>> > >
>> > > > via an index is probably flawed in the first place.
>> > >
>> > > Conversely, there are many people who seem to want Julia to treat
>> scalars
>> > > and
>> > > 1-vectors indistinguishably (ala Matlab).
>> > >
>> > > For what it's worth, here's a (contrived) example to justify the
>> current
>> > > behavior:
>> > >
>> > > function sum_over_dims(A, dims)
>> > >
>> > > for d in dims
>> > >
>> > > A = sum(A, d)
>> > >
>> > > end
>> > > A
>> > >
>> > > end
>> > >
>> > > sum_over_dims(A, [2,3])
>> > > sum_over_dims(A, 2)
>> > >
>> > > Why should I write sum_over_dims(A, [2]) in the latter case?
>> > >
>> > > Best,
>> > > --Tim
>>
>>
Re: [julia-users] getindex for a real number
It's a start :)
https://github.com/EricForgy/strict.jl
julia> using strict
julia> a = 5
5
julia> a[1]
ERROR: MethodError: `getindex` has no method matching getindex(::Type{Number
}, ::Type{Integer})
Closest candidates are:
getindex(::Type{T}, ::Any...)
getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{T})
getindex{T<:Union{Char,Number}}(::Type{T<:Union{Char,Number}}, ::Range{T},
::Range{T}...)
in getindex at C:\Users\Eric Forgy\.julia\v0.4\strict\src\strict.jl:4
On Wednesday, December 2, 2015 at 11:02:45 AM UTC+8, Tim Holy wrote:
>
> Likewise, I do see why this is a little troublesome. It's annoying when
> you
> mean to write `for i = 1:n` but accidentally write `for i = n`; it's not
> always an easy bug to find.
>
> --Tim
>
> On Tuesday, December 01, 2015 06:38:46 PM Eric Forgy wrote:
> > It bugs me, but only a little, so I won't lose sleep over it :)
> >
> > Then again, I wish Julia had a "strict" mode. In strict mode, the
> language
> > would be more pure mathematically, e.g. scalars have no indices, the
> > transpose of a vector is a covector, etc. This bit me recently because
> if T
> > <: U, then Array{T} is NOT <: Array{U} although as, sub-modules
> > , Tmodule <:
> Umodule.
> >
> > Then again, as I'm learning, if we want Julia to do something bad
> enough,
> > e.g. have a "strict" mode, we can have it. For example, I could write a
> > package "strict.jl" where
> >
> > using strict
> >
> > would kill Base.getindex(::Number) and things like that. That could be
> cool
> >
> > :)
> >
> > On Wednesday, December 2, 2015 at 9:38:50 AM UTC+8, Tim Holy wrote:
> > > On Tuesday, December 01, 2015 03:19:33 PM Eric Forgy wrote:
> > > > A scalar is distinct from a vector so size(a) = () makes sense.
> getindex
> > >
> > > for
> > >
> > > > a scalar does not make sense and should probably be removed on the
> > >
> > > grounds
> > >
> > > > of mathematical elegance :) Any code that depends on referencing a
> > >
> > > scalar
> > >
> > > > via an index is probably flawed in the first place.
> > >
> > > Conversely, there are many people who seem to want Julia to treat
> scalars
> > > and
> > > 1-vectors indistinguishably (ala Matlab).
> > >
> > > For what it's worth, here's a (contrived) example to justify the
> current
> > > behavior:
> > >
> > > function sum_over_dims(A, dims)
> > >
> > > for d in dims
> > >
> > > A = sum(A, d)
> > >
> > > end
> > > A
> > >
> > > end
> > >
> > > sum_over_dims(A, [2,3])
> > > sum_over_dims(A, 2)
> > >
> > > Why should I write sum_over_dims(A, [2]) in the latter case?
> > >
> > > Best,
> > > --Tim
>
>
Re: [julia-users] getindex for a real number
Likewise, I do see why this is a little troublesome. It's annoying when you
mean to write `for i = 1:n` but accidentally write `for i = n`; it's not
always an easy bug to find.
--Tim
On Tuesday, December 01, 2015 06:38:46 PM Eric Forgy wrote:
> It bugs me, but only a little, so I won't lose sleep over it :)
>
> Then again, I wish Julia had a "strict" mode. In strict mode, the language
> would be more pure mathematically, e.g. scalars have no indices, the
> transpose of a vector is a covector, etc. This bit me recently because if T
> <: U, then Array{T} is NOT <: Array{U} although as, sub-modules
> , Tmodule <: Umodule.
>
> Then again, as I'm learning, if we want Julia to do something bad enough,
> e.g. have a "strict" mode, we can have it. For example, I could write a
> package "strict.jl" where
>
> using strict
>
> would kill Base.getindex(::Number) and things like that. That could be cool
>
> :)
>
> On Wednesday, December 2, 2015 at 9:38:50 AM UTC+8, Tim Holy wrote:
> > On Tuesday, December 01, 2015 03:19:33 PM Eric Forgy wrote:
> > > A scalar is distinct from a vector so size(a) = () makes sense. getindex
> >
> > for
> >
> > > a scalar does not make sense and should probably be removed on the
> >
> > grounds
> >
> > > of mathematical elegance :) Any code that depends on referencing a
> >
> > scalar
> >
> > > via an index is probably flawed in the first place.
> >
> > Conversely, there are many people who seem to want Julia to treat scalars
> > and
> > 1-vectors indistinguishably (ala Matlab).
> >
> > For what it's worth, here's a (contrived) example to justify the current
> > behavior:
> >
> > function sum_over_dims(A, dims)
> >
> > for d in dims
> >
> > A = sum(A, d)
> >
> > end
> > A
> >
> > end
> >
> > sum_over_dims(A, [2,3])
> > sum_over_dims(A, 2)
> >
> > Why should I write sum_over_dims(A, [2]) in the latter case?
> >
> > Best,
> > --Tim
Re: [julia-users] getindex for a real number
It bugs me, but only a little, so I won't lose sleep over it :)
Then again, I wish Julia had a "strict" mode. In strict mode, the language
would be more pure mathematically, e.g. scalars have no indices, the
transpose of a vector is a covector, etc. This bit me recently because if T
<: U, then Array{T} is NOT <: Array{U} although as, sub-modules
, Tmodule <: Umodule.
Then again, as I'm learning, if we want Julia to do something bad enough,
e.g. have a "strict" mode, we can have it. For example, I could write a
package "strict.jl" where
using strict
would kill Base.getindex(::Number) and things like that. That could be cool
:)
On Wednesday, December 2, 2015 at 9:38:50 AM UTC+8, Tim Holy wrote:
>
> On Tuesday, December 01, 2015 03:19:33 PM Eric Forgy wrote:
> > A scalar is distinct from a vector so size(a) = () makes sense. getindex
> for
> > a scalar does not make sense and should probably be removed on the
> grounds
> > of mathematical elegance :) Any code that depends on referencing a
> scalar
> > via an index is probably flawed in the first place.
>
> Conversely, there are many people who seem to want Julia to treat scalars
> and
> 1-vectors indistinguishably (ala Matlab).
>
> For what it's worth, here's a (contrived) example to justify the current
> behavior:
>
> function sum_over_dims(A, dims)
> for d in dims
> A = sum(A, d)
> end
> A
> end
>
> sum_over_dims(A, [2,3])
> sum_over_dims(A, 2)
>
> Why should I write sum_over_dims(A, [2]) in the latter case?
>
> Best,
> --Tim
>
>
Re: [julia-users] getindex for a real number
On Tuesday, December 01, 2015 03:19:33 PM Eric Forgy wrote: > A scalar is distinct from a vector so size(a) = () makes sense. getindex for > a scalar does not make sense and should probably be removed on the grounds > of mathematical elegance :) Any code that depends on referencing a scalar > via an index is probably flawed in the first place. Conversely, there are many people who seem to want Julia to treat scalars and 1-vectors indistinguishably (ala Matlab). For what it's worth, here's a (contrived) example to justify the current behavior: function sum_over_dims(A, dims) for d in dims A = sum(A, d) end A end sum_over_dims(A, [2,3]) sum_over_dims(A, 2) Why should I write sum_over_dims(A, [2]) in the latter case? Best, --Tim
Re: [julia-users] getindex for a real number
A scalar is distinct from a vector so size(a) = () makes sense. getindex for a scalar does not make sense and should probably be removed on the grounds of mathematical elegance :) Any code that depends on referencing a scalar via an index is probably flawed in the first place.
Re: [julia-users] getindex for a real number
Then I remember sth else. Most probably I have tried ones(size(a)) in Julia when `a` is a real number, and it does not do what the same command does in Matlab. There are some other indexing that work in Julia, e.g., a=1.0 a[1] a[1,1] a[1,1,1,1] a[end,end,end] all return 1.0. But if I say size(a) it returns (). Or if I say b=ones(1,1,1) then again b[1,1,1,1,1] or b[1,1,:,1] return 1.0, but b[1,1,1,:] throws an error. On Monday, November 30, 2015 at 10:47:39 PM UTC+1, Mauro wrote: > > Your welcome. Your recollection is wrong though, this was possible in > 0.2 and 0.3; I just checked (I don't have a 0.1 build but maybe I > should...). > > On Mon, 2015-11-30 at 22:25, Ehsan Eftekhari > wrote: > > Thanks Mauro. As you predicted, my question was about the reasoning > behind > > this behavior. If I remember correctly, it was not possible to use > getindex > > on real numbers in Julia 0.3. I was not sure why it is added to 0.4. > > > > On Monday, November 30, 2015 at 10:01:54 AM UTC+1, Mauro wrote: > >> > >> As it says, there is no method for it. You could add it yourself: > >> > >> Base.getindex(f::Float64, ::Colon) = f # or whatever you like > >> > >> However, the philosophical question is: should you be allowed to index > >> into a float (or int)? Julia usually puts convenience before > strictness > >> and allows this. I'm not sure there are technical reasons to draw the > >> line at `1[:]`. If no-one gives one, you could open a pull request > >> (probably should add a method for `q[1:end]` too). > >> > >> On Mon, 2015-11-30 at 09:32, Ehsan Eftekhari >> > wrote: > >> > I have a question about this behaviour of getindex in Julia: > >> > > >> > if I say > >> > > >> > julia> a=1.0 > >> > 1.0 > >> > > >> > then > >> > > >> > julia> a[1] > >> > 1.0 > >> > > >> > julia> a[end] > >> > 1.0 > >> > > >> > but > >> > > >> > julia> a[:] > >> > ERROR: MethodError: `getindex` has no method matching > >> getindex(::Float64, > >> > ::Colon) > >> > Closest candidates are: > >> > getindex(::Number) > >> > getindex(::Number, ::Integer) > >> > getindex(::Number, ::Integer...) > >> > ... > >> > > >> > Why the last one does not work? > >> >
Re: [julia-users] getindex for a real number
Your welcome. Your recollection is wrong though, this was possible in 0.2 and 0.3; I just checked (I don't have a 0.1 build but maybe I should...). On Mon, 2015-11-30 at 22:25, Ehsan Eftekhari wrote: > Thanks Mauro. As you predicted, my question was about the reasoning behind > this behavior. If I remember correctly, it was not possible to use getindex > on real numbers in Julia 0.3. I was not sure why it is added to 0.4. > > On Monday, November 30, 2015 at 10:01:54 AM UTC+1, Mauro wrote: >> >> As it says, there is no method for it. You could add it yourself: >> >> Base.getindex(f::Float64, ::Colon) = f # or whatever you like >> >> However, the philosophical question is: should you be allowed to index >> into a float (or int)? Julia usually puts convenience before strictness >> and allows this. I'm not sure there are technical reasons to draw the >> line at `1[:]`. If no-one gives one, you could open a pull request >> (probably should add a method for `q[1:end]` too). >> >> On Mon, 2015-11-30 at 09:32, Ehsan Eftekhari > > wrote: >> > I have a question about this behaviour of getindex in Julia: >> > >> > if I say >> > >> > julia> a=1.0 >> > 1.0 >> > >> > then >> > >> > julia> a[1] >> > 1.0 >> > >> > julia> a[end] >> > 1.0 >> > >> > but >> > >> > julia> a[:] >> > ERROR: MethodError: `getindex` has no method matching >> getindex(::Float64, >> > ::Colon) >> > Closest candidates are: >> > getindex(::Number) >> > getindex(::Number, ::Integer) >> > getindex(::Number, ::Integer...) >> > ... >> > >> > Why the last one does not work? >>
Re: [julia-users] getindex for a real number
Thanks Mauro. As you predicted, my question was about the reasoning behind this behavior. If I remember correctly, it was not possible to use getindex on real numbers in Julia 0.3. I was not sure why it is added to 0.4. On Monday, November 30, 2015 at 10:01:54 AM UTC+1, Mauro wrote: > > As it says, there is no method for it. You could add it yourself: > > Base.getindex(f::Float64, ::Colon) = f # or whatever you like > > However, the philosophical question is: should you be allowed to index > into a float (or int)? Julia usually puts convenience before strictness > and allows this. I'm not sure there are technical reasons to draw the > line at `1[:]`. If no-one gives one, you could open a pull request > (probably should add a method for `q[1:end]` too). > > On Mon, 2015-11-30 at 09:32, Ehsan Eftekhari > wrote: > > I have a question about this behaviour of getindex in Julia: > > > > if I say > > > > julia> a=1.0 > > 1.0 > > > > then > > > > julia> a[1] > > 1.0 > > > > julia> a[end] > > 1.0 > > > > but > > > > julia> a[:] > > ERROR: MethodError: `getindex` has no method matching > getindex(::Float64, > > ::Colon) > > Closest candidates are: > > getindex(::Number) > > getindex(::Number, ::Integer) > > getindex(::Number, ::Integer...) > > ... > > > > Why the last one does not work? >
Re: [julia-users] getindex for a real number
As it says, there is no method for it. You could add it yourself: Base.getindex(f::Float64, ::Colon) = f # or whatever you like However, the philosophical question is: should you be allowed to index into a float (or int)? Julia usually puts convenience before strictness and allows this. I'm not sure there are technical reasons to draw the line at `1[:]`. If no-one gives one, you could open a pull request (probably should add a method for `q[1:end]` too). On Mon, 2015-11-30 at 09:32, Ehsan Eftekhari wrote: > I have a question about this behaviour of getindex in Julia: > > if I say > > julia> a=1.0 > 1.0 > > then > > julia> a[1] > 1.0 > > julia> a[end] > 1.0 > > but > > julia> a[:] > ERROR: MethodError: `getindex` has no method matching getindex(::Float64, > ::Colon) > Closest candidates are: > getindex(::Number) > getindex(::Number, ::Integer) > getindex(::Number, ::Integer...) > ... > > Why the last one does not work?
[julia-users] getindex for a real number
I have a question about this behaviour of getindex in Julia: if I say julia> a=1.0 1.0 then julia> a[1] 1.0 julia> a[end] 1.0 but julia> a[:] ERROR: MethodError: `getindex` has no method matching getindex(::Float64, ::Colon) Closest candidates are: getindex(::Number) getindex(::Number, ::Integer) getindex(::Number, ::Integer...) ... Why the last one does not work?
