search

JuliaSmoothOptimizers / NLPModels.jl

Data Structures for Optimization Models
Other
173 stars 35 forks source link

Regarding the update to 0.7 #125

Closed abelsiqueira closed 5 years ago

abelsiqueira commented 6 years ago

A possibility to handle 0.7 now, would be to separate JuMP and MPB related things to a separate package (NLPJuMPModel?) and keep it 0.6 while MOI is not released, then this can be updated.

Also, should we be worried about handling both MPB and MOI, and just drop MPB when the time comes?

vepiteski commented 6 years ago

As a user/developer, I am puzzled. I wish to upgrade to 0.7 asap. On the other hand, most of my code relies on JuMP/MPB.

Personally, I prefer to defer upgrading until the dust has settled down... which means living off-beat with the Julia community, unfortunately.

Great to have seen a spectacular announcement of 1.0, but Julia without its numerous packages is nothing, so the real upgrade will be completed no sooner than when /all/ of the packages I rely upon will be 0.7/1.0 compatible.

Best,

JPD

P.S. My post does not help deciding for the upgrade strategy of JuliaSmoothOptimizers, I am afraid!

Le 2018-08-15 à 20:43, Abel Siqueira a écrit :

  • JuMP-0.18/MPB will be supported only on 0.6;
  • JuMP-0.19/MOI will be supported on both 0.6 and 0.7, but it's not released yet. What should be our update strategy?

A possibility to handle 0.7 now, would be to separate JuMP and MPB related things to a separate package (NLPJuMPModel?) and keep it 0.6 while MOI is not released, then this can be updated.

Also, should we be worried about handling both MPB and MOI, and just drop MPB when the time comes?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/JuliaSmoothOptimizers/NLPModels.jl/issues/125, or mute the thread https://github.com/notifications/unsubscribe-auth/AHy_G-nlws8BUVjQr9sisaB-aG3ix86Qks5uRMAggaJpZM4V_Bgg.

dpo commented 6 years ago

What is the rush to upgrade to 0.7? Maybe we just need to stay with 0.6 for a while?!

I suppose we could extract JumpNLPModel (or whatever) from NLPModels. It would be consistent with what we did for CUTEst and AMPL models.

abelsiqueira commented 6 years ago

Sorry, I should say support 0.7 instead of update to 0.7. I'm looking more on having CUTEst passing on 0.7, but I'll probably stick with 0.6 for a few months.

I'll work on separating it then.