[SciPy-dev] [scikits] openopt SVN instable for the moment

[dmitrey]

Matthieu Brucher wrote:
>  
>
>     P.S. OO installation page says extracting OO from svn is recommended
>     way. So users should be informed of the issue, not just scipy-dev.
>
>
>
> That's what packages are made for. When the package is in a stable 
> state, one can think about modifying files so that the module is 
> enhanced. This fix waited far too long. And besides, I think that the 
> code I provided in the scikit should be available for people who want 
> it (generic framework that you hid in the solvers).

In the letter from 09/12/2007 I had proposed you to install your 
optimizers to same path OpenOpt is installed, i.e. using from 
scikits.genericopt import ...
You had answered: "Do what you think is the best, I don't mind using 
from openopt... import optimizers"


> I once started a similar fix and you rolled back everything.

I had got to know about those changes when Nils Wagner had informed me 
that latest OO from subversion just don't work.

> I'm sorry, but you are not the only one that spent days, weeks, to 
> create the package. I spent hours trying to do something you said you 
> would do months ago (this fix). But sometimes, decisions have to be 
> made, because other code depend on it. I already waited too long to 
> fix this design (this was one of the reasons I did not add new stuff 
> to the package, as I had no way of advertizing it to the outside world).
Do you realize what any project will look like if anyone will implement 
his own changes wrt his own needs only, w/o even informing others?
>  
>
>     On
>     the other hand, I prefer to check all /examples/*.py pass ok BEFORE
>     committing anything to svn.
>
>
> That cannot be always the case. Besides, since your last release, I 
> don't think you commited code,
I have committed some since v 0.17 (changes to ralg + bugfix to 
ALGENCAN), and I'm just waiting for GSoC decision that have to be 
announced 11 April. Since I had mentioned my intentions of participating 
in GSoC in my blog that had been connected to scipy-planet by someone, I 
guess you are aware of this. BTW anyone can have some holidays during year.

> so people who want to use the last stable realease can do so. And 
> people who want to use the full power of the package can do so now.
I still don't think some days of my not-committing give rights to those 
abstract people you have mentioned to do whatever they want with OO code.

>
> Sorry for the harsh criticism, but I 'm thinking about this for 
> several months.
>
> Matthieu
And after those several months you can't wait these 2 days before GSoC 
results?

D.

>
>
> 2008/4/9, dmitrey <dmitrey.kroshko at scipy.org 
> <mailto:dmitrey.kroshko at scipy.org>>:
>
>     So now nlp_ALGENCAN.py and milp_1.py just don't work, mb some
>     others as
>     well.
>
>
>
> Perhaps with the erros, I could be able to find out what is happening ?
Traceback (most recent call last):
  File "nlp_ALGENCAN.py", line 112, in <module>
    r = p.solve('ALGENCAN')
  File 
"/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/BaseProblem.py", 
line 231, in solve
    return runProbSolver(self, solvers, *args, **kwargs)
  File 
"/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/runProbSolver.py", 
line 67, in runProbSolver
    solverClass =  getattr(my_import(available_solvers[solver_str]), 
solver_str)
  File 
"/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/runProbSolver.py", 
line 50, in my_import
    mod = __import__(name)
  File 
"/usr/lib/python2.5/site-packages/scikits/openopt/solvers/BrasilOpt/ALGENCAN_oo.py", 
line 5, in <module>
    from scikits.openopt.setDefaultIterFuncs import SMALL_DF
ImportError: No module named setDefaultIterFuncs


Traceback (most recent call last):
  File "milp_1.py", line 17, in <module>
    r = p.solve('lpSolve')
  File 
"/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/BaseProblem.py", 
line 231, in solve
    return runProbSolver(self, solvers, *args, **kwargs)
  File 
"/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/runProbSolver.py", 
line 199, in runProbSolver
    solver(p)
  File 
"/usr/lib/python2.5/site-packages/scikits/openopt/solvers/lp_solve/lpSolve_oo.py", 
line 26, in __solver__
    [obj, x_opt, duals] = lps(List(f.flatten()), List(p.Awhole), 
List(p.bwhole.flatten()), List(p.dwhole.flatten()), \
NameError: global name 'lps' is not defined