Warning when re-solving a model

[userdev163]

Im sorry but it is very difficult to put code examples for my issue. I'm using ortools 9.11. I have the following workflow in my project:
I create a scheduling model and add constraints (no overlap) until the problem is not solvable.
Then I reuse the last feasible model that I saved (using a DeepCopy) and I solve this model again adding noOverlap constraints from another source. This raises :
"Warning, an interval constraint was likely used without a corresponding linear constraint linking its start, size and end."
When I solve the model (cp_model) again via:
solver = cp_model.CpSolver()
solver.Solve(self.model)
I have tried to solve again adding no constraints and I have the same result.
My best guess for know is that I have a constraint with a start value at 0, which would be a problem when reloaded (when exported to file this start becomes empty)
What could I be doing wrong in this situation?

[sschnug]

Did you follow sat/docs/model.md -> the variable-mapping step (get_int_var_from_proto_index)?

[lperron]

In 9.11 creating an interval also adds the linear constraint that enforces the start +size==end

This has been removed.

Still I do not know why it fails.

Note that copying is tricky and error prone. I recommend rebuilding the model

[userdev163]

I have narrowed down my issue, it seems that the problem appears on my model because of a variable with a start = 0
When I'm giving an hint solution including this variable, the warning is generated. But when I skip this variable in the hint solution, no warning.
I have also noticed that the solveur seems to remove this start = 0 and replace it with {} (when i print the model)
I think this may be related.

[lperron]

printing: this is the default behavior. 0 is the default value and is not printed. Laurent Perron | Operations Research | ***@***.*** | (33) 1 42 68 53 00 Le mar. 26 nov. 2024 à 09:47, userdev163 ***@***.***> a écrit :

…

I have narrowed down my issue, it seems that the problem appears on my model because of a variable with a start = 0 When I'm giving an hint solution including this variable, the warning is generated. But when I skip this variable in the hint solution, no warning. I have also noticed that the solveur seems to remove this start = 0 and replace it with {} (when i print the model) I think this may be related. — Reply to this email directly, view it on GitHub <#4448 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACUPL3OHYUB2Q6A2N4LMZJL2CQ7SNAVCNFSM6AAAAABSHHBYWKVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCMZYGE4DENI> . You are receiving this because you commented.Message ID: ***@***.***>