[Coin-discuss] Compilation error

[Matthew Saltzman]

On Thu, 13 Apr 2006, Brady Hunsaker wrote:

> The error below in the unit test for OsiGlpk is a (recently) known
> error.  This report is extra motivation for me to do something about it.
>
> The problem is that GLPK fails to resolve after removing a constraint.
> So it's not really the interface so much as a fundamental capability of
> GLPK.  I need to determine whether it's better to create a workaround in
> the interface or try to change GLPK upstream so that it can handle the
> situation.  It's possible that a little more investigating will reveal
> something else, though.
>
> It's on my to-do list.  We don't have a bug-tracking system up for OSI
> yet, do we?

Not quite, but we are close.  Asking helps the motivation factor here too 
8^).

>
> Brady
>
> Matthew Saltzman wrote:
>> On Thu, 13 Apr 2006, Cezik, Tolga wrote:
>>
>>> Thank you Ted!
>>>
>>> I thought SYMPHONY-5.0 was newer but now I realize 5.1 was the one to
>>> use. Everything seems OK. However tests are usually giving errors here
>>> and there. For instance (this is Osi related, i.e. the test in
>>> COIN/Osi/.)
>>> ----
>>> Coin0008I P0033 read with 0 errors
>>> spx_simplex: initial basis is invalid
>>> unitTest: OsiGlpkSolverInterfaceTest.cpp:975: void
>>> OsiGlpkSolverInterfaceUnitTest(const std::string&, const
>>> std::string&): Assertion `eq(objValue,2520.57)' failed.
>>> zsh: abort      ./unitTest -testOsiSolverInterface
>>> ----
>>> And if I comment that line out, then
>>> ----
>>> Coin0008I 25FV47 read with 0 errors
>>> eq 0 1.79769e+308 1e+300
>>> unitTest: OsiSolverInterfaceTest.cpp:1324: void
>>> OsiSolverInterfaceMpsUnitTest(const std::vector<OsiSolverInterface*,
>>> std::allocator<OsiSolverInterface*> >&, const std::string&): Assertion
>>> `equivalentVectors(vecSiP[i-1],vecSiP[i], 1.e-10,
>>> vecSiP[i-1]->getColUpper(),vecSiP[i ]->getColUpper(),vecSiP[i
>>> ]->getNumCols() )' failed.
>>> zsh: abort      ./unitTest -testOsiSolverInterface
>>> ----
>>> I am not too concerned of these errors, should I?
>>
>>
>> Well, the unit tests *should* pass.  That's why they're unit tests 8^).
>>
>> You're using GLPK as a solver?  What version?  What OS/compiler and
>> versions?
>>
>>>
>>> --tolga
>>>
>>>
>>>
>>>
>>
>
> _______________________________________________
> Coin-discuss mailing list
> Coin-discuss at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-discuss
>

-- 
 		Matthew Saltzman

Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs