I believe that Andreas did this to make it consistent with the dense case.
You want to do:
julia> ones(10)'*sprand(10,10,0.1)
1x10 Array{Float64,2}:
0.871294 0.820477 0.969049 2.02607 … 0.74965 0.989042 0.0 0.991852
Both of these do not work now:
julia> ones(10)*rand(10,10)
ERROR: DimensionMismatch("A has dimensions (10,1) but B has dimensions (10,
10)")
in gemm_wrapper! at linalg/matmul.jl:295
in * at linalg/matmul.jl:70
julia> ones(10)*sprand(10,10,0.1)
ERROR: DimensionMismatch("")
in * at sparse/linalg.jl:104
in * at linalg/matmul.jl:70
-viral
On Tuesday, June 30, 2015 at 1:07:33 PM UTC-4, Seth wrote:
>
> Following up again: it appears that x::Vector * A::SparseMatrixCSC was
> removed via a commit last week.
> <https://github.com/JuliaLang/julia/commit/d914fb34a75d1c0246195665ad63a6b01a733913>
> Is
> anyone aware of a workaround?
>
>
>
> On Tuesday, June 30, 2015 at 11:06:05 AM UTC-5, Seth wrote:
>>
>> Following up, it's not with @test, it's with the pagerank code itself:
>>
>> ERROR: LoadError: LoadError: DimensionMismatch("")
>>
>> in * at sparse/linalg.jl:104
>>
>> in * at linalg/matmul.jl:70
>>
>> in pagerank at
>> /home/travis/.julia/v0.4/LightGraphs/src/centrality/pagerank.jl:18
>>
>>
>> This works in 0.4.0-dev+5008 but is broken in nightly.
>>
>>
>> On Tuesday, June 30, 2015 at 10:31:58 AM UTC-5, Seth wrote:
>>>
>>> Trying to get Travis to behave, but I'm running into a new (for me)
>>> error message. Does anyone have any ideas? That should be rounding a
>>> Float64 to three decimal places. It works on 0.4.0-dev+5008.
>>>
>>> The line is
>>>
>>> @test round(pagerank(g5)[3],3) == 0.318
>>>
>>> (I don't know where those extra parens (in the output below) are coming
>>> from, but on my system it works. It's just not working on Travis.)
>>>
>>>
>>>
>>> ERROR: LoadError: LoadError: test error in expression:
>>> round((pagerank(g5))[3],3) == 0.318
>>>
>>> DimensionMismatch("")
>>>
>>>
