Update MOI_callbacks.jl

[odow]

x-ref: https://discourse.julialang.org/t/jump-0-20-1-possible-bug-in-cplex-when-using-generic-lazy-callback/35425

[codecov-io]

Codecov Report

Merging #286 into master will increase coverage by 0.04%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #286      +/-   ##
==========================================
+ Coverage   69.43%   69.47%   +0.04%     
==========================================
  Files          15       15              
  Lines        2460     2457       -3     
==========================================
- Hits         1708     1707       -1     
+ Misses        752      750       -2

Impacted Files	Coverage Δ
src/MOI/MOI_callbacks.jl	89.61% <100%> (+2.11%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1e24a71...19c82b9. Read the comment docs.

[ryancorywright]

Thanks for looking into this. I applied the patch and reran the code mentioned in the discourse page: CPLEX doesn't throw an error any more, but it rejects every feasible solution and eventually says that the problem is infeasible.

From the docs https://www.ibm.com/support/knowledgecenter/SSSA5P_12.10.0/ilog.odms.cplex.help/refcallablelibrary/macros/CPX_CALLBACKCONTEXT_RELAXATION.html

"CPLEX invokes the generic callback in this context when it has found a relaxed solution available. The relaxed solution is usually not integer feasible. It can, for example, be the solution to a node LP (but can also come from another place)."

[odow]

On this branch we only call the lazy constraint from CPX_CALLBACKCONTEXT_CANDIDATE:
https://github.com/JuliaOpt/CPLEX.jl/blob/19c82b917b173189b45a4818331d53dee86a0e8d/src/MOI/MOI_callbacks.jl#L81-L89

Please provide a reproducible example.