tests failing on julia 0.6 #192
Comments
|
I generally find I'm unable to compile Julia from source; but I'll start
debugging this once there's a compiled prerelease. If anyone else wants to
figure out what's going on in the meantime, please do!
…On Sun, Mar 19, 2017 at 3:46 PM, Miles Lubin ***@***.***> wrote:
In addition to deprecation warnings, I'm getting the following failure
when running Convex.jl's tests on the latest julia 0.6 prerelease:
> index atom
ERROR: LoadError: LoadError: LoadError: MethodError: no method matching getindex(::Convex.Variable, ::CartesianIndex{2})
Stacktrace:
[1] collect(::Base.Generator{Base.LogicalIndex{CartesianIndex{2},Array{Bool,2}},Convex.##34#35{Convex.Variable}}) at ./array.jl:418
[2] getindex(::Convex.Variable, ::Array{Bool,2}) at /home/mlubin/.julia/v0.6/Convex/src/atoms/affine/index.jl:88
[3] (::##120#385)() at /home/mlubin/.julia/v0.6/Convex/test/test_affine.jl:136
[4] context(::##120#385, ::String) at /home/mlubin/.julia/v0.6/FactCheck/src/FactCheck.jl:475
[5] (::##37#302)() at /home/mlubin/.julia/v0.6/Convex/test/test_affine.jl:126
[6] facts(::##37#302, ::String) at /home/mlubin/.julia/v0.6/FactCheck/src/FactCheck.jl:449
[7] include_from_node1(::String) at ./loading.jl:539
[8] include(::String) at ./sysimg.jl:14
[9] macro expansion at /home/mlubin/.julia/v0.6/Convex/test/runtests_single_solver.jl:22 [inlined]
[10] anonymous at ./<missing>:?
[11] include_from_node1(::String) at ./loading.jl:539
[12] include(::String) at ./sysimg.jl:14
[13] macro expansion at /home/mlubin/.julia/v0.6/Convex/test/runtests.jl:36 [inlined]
[14] anonymous at ./<missing>:?
[15] include_from_node1(::String) at ./loading.jl:539
[16] include(::String) at ./sysimg.jl:14
[17] process_options(::Base.JLOptions) at ./client.jl:305
[18] _start() at ./client.jl:371
while loading /home/mlubin/.julia/v0.6/Convex/test/test_affine.jl, in expression starting on line 6
while loading /home/mlubin/.julia/v0.6/Convex/test/runtests_single_solver.jl, in expression starting on line 20
while loading /home/mlubin/.julia/v0.6/Convex/test/runtests.jl, in expression starting on line 32
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#192>, or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAyp9GcXMfq3a3yLkKKM9rhqCqDPpyGSks5rnYYZgaJpZM4Mh2KX>
.
--
Madeleine Udell
Assistant Professor, Operations Research and Information Engineering
Cornell University
https://people.orie.cornell.edu/mru8/
(415) 729-4115
|
|
There is a compiled prerelease: https://discourse.julialang.org/t/julia-0-6-alpha-released-please-update-your-projects/2571 |
|
Ok, here's an update I could use help on. Broadcasting in Julia v0.6 is automatic; one no longer defines methods However, for us that's a problem, since if x and y are both matrix variables, The simplest way to fix this problem would be to redefine broadcast, which I have so far been unable to do. See the branch ready-for-v0.6. I'd appreciate help here. The nicest way (I think), which requires a significant rewrite of the code base, would be to make all matrix/vector variables into arrays of variables, so .* would automatically do the right thing. It's not clear how to do this without introducing significant overhead, since we'd have to check convexity / do conic form transformations elementwise. |
|
Yes, the right thing is to redefine |
|
For a non-toy example that utilizes the full (We are planning to make |
|
Oh, I see, life is more complicated for you because you need to guarantee convexity, so it is not possible for you to extend |
|
Closed by ba00f76. Unfortunately, it required a breaking change to the syntax. |
In addition to deprecation warnings, I'm getting the following failure when running Convex.jl's tests on the latest julia 0.6 prerelease:
The text was updated successfully, but these errors were encountered: