Running @spawn solve(m) gives an stack overflow

[h3nnn4n]

When running @Spawn solve(m) on the code below a stack overflow is given.

using JuMP

m = Model()
@defVar(m, 0 <= x <= 2 )
@defVar(m, 0 <= y <= 30 )

@setObjective(m, Max, 5x + 3*y )
@addConstraint(m, 1x + 5y <= 3.0 )

ref = @spawn solve(m)

The error:

ERROR: stack overflow
 in resize! at ./array.jl:506
 in ensureroom at ./iobuffer.jl:182
 in write at ./iobuffer.jl:267
 in serialize at serialize.jl:89
 in serialize_type_data at serialize.jl:271
 in serialize at serialize.jl:289
 in serialize at serialize.jl:78
 in serialize_type_data at serialize.jl:278
 in serialize_type at serialize.jl:299
 in serialize at serialize.jl:318
 in serialize at serialize.jl:324 (repeats 3 times)
...
 in serialize at serialize.jl:134
 in serialize at serialize.jl:324 (repeats 4 times)
 in serialize at serialize.jl:134
 in serialize at serialize.jl:324 (repeats 3 times)
 in serialize at serialize.jl:78
 in serialize at serialize.jl:218
 in send_msg_ at multi.jl:172
 in remotecall at multi.jl:686
 in remotecall at multi.jl:690
 in spawn_somewhere at multi.jl:1276
 in include at ./boot.jl:245
 in include_from_node1 at loading.jl:128
 in process_options at ./client.jl:285
 in _start at ./client.jl:354
while loading /homenis/renan/julia_bnb/new/bnb.jl, in expression starting on line 10
WARNING: Forcibly interrupting busy workers
WARNING: Unable to terminate all workers

The only way around I found so far is to create the model on the worker where it will run, returning the model to work on later ends up on the same problem as above.

[mlubin]

JuMP models aren't serializable across julia processes. Even if they were, it wouldn't be the most compact or efficient way to transfer a problem instance.

[h3nnn4n]

So I cant @async solve on a couple of models?

[mlubin]

Currently no. Leaving open as a feature request.
If you'd like to take a stab at getting this working, we're happy to help out. I'm not entirely sure what the major technical challenges are.

[mlubin]

Just tested on Julia 0.5, and this example now works. There are still open issues with using JuMP within parallel code, however (#673).