[Coin-osi-devel] OSI testing

[Matthew Saltzman]

On Wed, 2008-01-02 at 10:32 -0800, Lou Hafer wrote:
> Matt,
> 
> 	I should have been more precise:  I'm talking about --test-solvers and
> --exclude-solvers as command line parameters for the OSI unitTest, not for the
> configure scripts.  A run-time solution rather than a compile-time solution.  We
> can do this within the bounds of the OSI project.

Ah, yes, I see.  OK how about:

./unitTest --include-solvers "..."
./unitTest --exclude-solvers "..."

Default if neither option given is all built solvers.  The include
option gets all solvers both built and listed and the exclude option
gets all solvers built but not listed.

> 
> 	COIN_SKIP_PROJECTS is pretty far-reaching; it affects the run_autotools
> script as well as configure scripts.  Often this is too far-reaching for me.  I
> may want to restrict the Osi unitTest to only OsiGlpk, but I don't want to have
> Clp and DyLP disappear from Cgl and Cbc as a side effect.

This seems like a good idea for other meta-projects as well (CGL in
particular).  If moving toward a unified test-harness architecture is a
worthy long-term goal, maybe more generic names for the options would be
in order?  --include-components/--exclude-components?

		Matt

> 
> 	There are a lot of good arguments for separating `project Foo is
> present, please configure and compile' from `please use project Foo when
> configuring and compiling project Bar'.  Many projects would benefit.  But this
> is a big change in the autotools architecture affecting many projects and needs
> wider discussion.
> 
> 						Lou
> 
> _______________________________________________
> Coin-osi-devel mailing list
> Coin-osi-devel at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-osi-devel
-- 
                Matthew Saltzman

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