don't force inlining for annotation(Inline=false)

[Vitalij Ruge]

see comments
#3785/9, #3785/11

we shouldn't force inline function with annotation(Inline=false)

possible steps:

1) added InlineType default
2) replace DAE.NO_INLINE() in e.g. Inline.forceInlineExp(inExp,(functions,{DAE.NORM_INLINE(),DAE.NO_INLINE()})
whit DAE.DEFAULT_INLINE()

[Vitalij Ruge]

fixed with ​PR568
and improved with ​PR582

from local testing -> we will get some impact for e.g.

ModelicaTest.Fluid.Dissipation.TestCases.PressureLoss.SeveralTestCases

let's see the next result from coverage :)

[Rüdiger Franke]

The time for ModelicaTest_trunk went down from 1931s yesterday to 1747s today. Great!

​https://test.openmodelica.org/libraries/ModelicaTest_trunk/BuildModelRecursive.html

[Lennart Ochel]

@vruge What do you mean by implicit inlining and to what kind of functions does it apply?

[Vitalij Ruge]

Replying to rfranke:

The time for ModelicaTest_trunk went down from 1931s yesterday to 1747s today.

I guess it fixes partial the jump 1800 to 1950 from ​solveSimpleEqnForSim.

imprecise approximate(without ​solveSimpleEqnForSim) it is 1782s to 1747s.

Great!

Thanks :)

[Francesco Casella]

Replying to rfranke:

The time for ModelicaTest_trunk went down from 1931s yesterday to 1747s today. Great!

​https://test.openmodelica.org/libraries/ModelicaTest_trunk/BuildModelRecursive.html

Better still, the time to build ModelicaTest.Fluid.Dissipation.TestCases.PressureLoss.Orifice went down from 264 s to 8.7 s (!). Also the simulation time has slightly improved, further confirming that library developer's advice should preferably be trusted :)

[Martin Sjölund]

Milestone deleted