[Coin-discuss] OsiCbcSolverInterface turn off log messages

[Ted Ralphs]

Why not just make the branchandbound() method of OsiClp invoke Cbc? This
is how most of the other interfaces work. OSI was never really intended
to be an interface to a code that is purely an IP solver (OsiSym is the
exception, though even OsiSym will pass the solve call through to the
underlying LP solver if no variables are declared integer). We don't
have separate solver interfaces for CPLEX's LP and IP solvers, for
instance. Why duplicate all the code for building and manipulating the
model? The only call that is specific to IP is branchandcound().

Cheers,

Ted

Matthew Saltzman wrote:
> On Fri, 9 Mar 2007, Tim Hultberg wrote:
> 
>> What?!
>>
>> I am very sorry to hear this. When I first wrote flopc++, I interfaced
>> directly with Cplex. Then I discovered Coin-OSI, and was able to change
>> solver easily without writing new interfaces each time. Now I use OsiCbc
>> as solver in the flopc++ example MIP models (or OsiClp, if its an LP),
>> such that they can be compiled and executed without using anything
>> external to Coin.
>>
>> I think it makes sense to keep a strong and up to date OsiCbc in Coin,
>> so maybee this is a good candidate for the help wanted list. (Or should
>> I use OsiSym instead?)
> 
> I agree that this should be a priority (for exactly this reason), and it
> would be great to have a brave volunteer to take it on.
> 
>>
>> Tim Hultberg
>>
>>
>>>>> John J Forrest <jjforre at us.ibm.com> 09/03/2007 16:14 >>>
>> By the way - the use of OsiCbc is deprecated - there is much more
>> functionality and performance using Cbc and OsiClp.
>>
>> John Forrest
>>
>>
>>
>>
>> _______________________________________________
>> Coin-discuss mailing list
>> Coin-discuss at list.coin-or.org
>> http://list.coin-or.org/mailman/listinfo/coin-discuss
>>
> 


-- 
Dr. Ted Ralphs
Associate Professor
Industrial and Systems Engineering
Lehigh University
(610)758-4784
tkralphs at lehigh.edu
www.lehigh.edu/~tkr2