John Siirola

`compute_variable_from_constraint` has an extra trick before the Newton solve: since most of the time is is actually called on an expression where `x` appears linearly, it does a spot-check assuming...

Oooo... that's really good. ...And i suppose not surprising - I am pretty sure the cost in the Newton solve is getting the symbolic derivative. Did you (or can you...

@michaelbynum: would it be easier to migrate coramin to use the `@declare_custom_block` from Pyomo? That implementation resolved a bunch of metaclass issues that the original IDAES implementation had.

This is particularly nasty. If you turn on debugging (`logging.getLogger('pyomo.repn').setLevel(logging.DEBUG)`), you will see: ``` DEBUG: NL presolve: substituting concentration[Cl] := -1.0*solid_flows[KCl] + -0.0 DEBUG: NL presolve: bounds fixed solid_flows[KCl] :=...

I have spent the last week debating (with myself [and losing]) what the "expected" behavior here should be. The current implementation dates back to the original NL writer, and followed...

@alma-walmsley: sorry this PR is dragging on -- most of the dev team has been out for the last 3 weeks (between holidays and an IDAES team meeting), and we...

The easiest address is [email protected] (that should reach a good number of us).

It is unfortunate that e1da345 was committed... that introduced a bunch of non-functional changes that are confusing review of this PR. For the future, we run (and test against) `black...

> This is my bad @jsiirola. In this case, would the best practice is to make a new PR on a clean branch with the changes to the import statement?...

@AnhTran01, would changing the import for GDX imply a minimum supported GAMS version (i.e., when did the import path change)?