Adding stubs for `init_state` #10560

amirroth · 2024-06-12T14:43:12Z

I created stubs for an init_state(EnergyPlusData &state) function that is the mirror of the clear_state() function and that is intended to allow us to incrementally serialize and simplify our input and data-mining architecture. I have not actually implemented any of that yet, so this is really just empty functions with one exception. I moved a call to GetFluidPropertiesData() into FluidPropertiesData::init_state(). There is nothing to see here. These are not the droids you're looking for. Move along.

Mac and Windows are green, Ubuntu is not building due to the Python version update.

amirroth

Bada bing, bada boom.

amirroth · 2024-06-12T14:43:50Z

src/EnergyPlus/AirLoopHVACDOAS.hh

@@ -183,9 +183,14 @@ struct AirLoopHVACDOASData : BaseGlobalStruct
    std::vector<AirLoopHVACDOAS::AirLoopDOAS> airloopDOAS;
    std::vector<AirLoopHVACDOAS::AirLoopMixer> airloopMixer;
    std::vector<AirLoopHVACDOAS::AirLoopSplitter> airloopSplitter;
+


You're going to see a lot of this.

amirroth · 2024-06-12T14:44:07Z

src/EnergyPlus/AirLoopHVACDOAS.hh

    void clear_state() override
    {
-        *this = AirLoopHVACDOASData();
+        new (this) AirLoopHVACDOASData();


And a little bit of this.

amirroth · 2024-06-12T14:44:33Z

src/EnergyPlus/Data/EnergyPlusData.cc

@@ -567,4 +567,262 @@ void EnergyPlusData::clear_state()
    this->files.ssz.close();
    this->files.zsz.close();
 }
+


Here is the uber-function.

amirroth · 2024-06-12T14:45:36Z

src/EnergyPlus/FluidProperties.hh

@@ -662,6 +662,12 @@ struct FluidPropertiesData : BaseGlobalStruct
    std::array<FluidProperties::cached_tsh, FluidProperties::t_sh_cache_size> cached_t_sh;
 #endif

+    void init_state(EnergyPlusData &state) override


The only non-empty init_state function thus far.

I noticed that FluidProperties.cc did not change which means these can also be deleted everywhere, as well as the variable state.dataFluidProps->GetInput.

if (state.dataFluidProps->GetInput) { GetFluidPropertiesData(state); state.dataFluidProps->GetInput = false; }

I'm keeping this pretty minimal for now (strange, I know). I'm going to rework the FluidProperties API and see the effects downstream before propagating this, but I will do that in the following PR.

amirroth · 2024-06-12T14:46:18Z

src/EnergyPlus/SimulationManager.cc

@@ -209,6 +209,8 @@ namespace SimulationManager {
             state.dataInputProcessing->inputProcessor->getNumObjectsFound(state, "RunPeriod:CustomRange") > 0 || state.dataSysVars->FullAnnualRun);
        state.dataErrTracking->AskForConnectionsReport = false; // set to false until sizing is finished

+        state.init_state(state);


I decided to plant the call to the uber-function here. Is there a better place for it?

It appears this allows calling all getInput functions before the simulation gets started. And yes, order of calls in EnergyPlusData::init_state is critically important. This will replace the isInputObjectUsed function down at line 229 which is also trying to get information prior to the simulation starting. So somewhere between line 212 and 229 seems appropriate to me. I think at the very least that this call should be moved down below OpenOutputFiles at line 214. Or there may be a need to reorganize calls between lines 214 and 229 where some of those lines could be moved to init_state or reorganized locally here depending on what each function does.

If I'm following correctly, ideally, we want to split up getInput into two parts: read and cross-check. e.g. Read all of the inputs but don't do any checking at first (e.g. checking for a zone name). Then the input order won't be (or could be less of) an issue.

Since getInputs write to output files (err for example) this call should be moved below OpenOutputFiles before merging. getFluidProperties does call Show*Error so the err file needs to be open first.

It appears this allows calling all getInput functions before the simulation gets started. And yes, order of calls in EnergyPlusData::init_state is critically important. This will replace the isInputObjectUsed function down at line 229 which is also trying to get information prior to the simulation starting. So somewhere between line 212 and 229 seems appropriate to me. I think at the very least that this call should be moved down below OpenOutputFiles at line 214. Or there may be a need to reorganize calls between lines 214 and 229 where some of those lines could be moved to init_state or reorganized locally here depending on what each function does.

Okay, I can move this wherever and we can keep moving it if needed. For now, it's essentially an empty function.

Since getInputs write to output files (err for example) this call should be moved below OpenOutputFiles before merging. getFluidProperties does call Show*Error so the err file needs to be open first.

Did you already do this?

No, but it does need to move. I'm not sure about the other calls below but that will show itself, as well as order dependence, once things start moving to the new init_state function.

OK, I just checked. OpenOutputFiles opens eso, eio, mtr, and bnd. I think the bnd (node list) needs to be open before calling getInput. The err file gets opened in initializeEnergyPlus and initializeAsLibrary which seems it's already open by this point. So this line still needs to move after OpenOutputFiles.

I moved it, although it looked like it was already moved for some reason (that's why I thought you did it). I'm going to make this transition over time, this just lets me get started.

Regarding: "split up getInput into two parts: read and cross-check" I think if we want to read inputs early we would want to know that data is valid and error checked so that there is no need for other checks elsewhere (i.e., if (zoneNum > 0), etc.). Just suggesting this does not move too fast other than trying to get a good order for getInput calls. Once that effort is well under way we could make decisions like that.

Myoldmopar · 2024-06-12T14:50:34Z

(Applied Clang Format and pulled develop so Ubuntu would be happy)

rraustad

Once these unrelated diffs are corrected and this init_state call in SimulationManager is moved this can merge.

rraustad · 2024-06-13T02:57:12Z

Myoldmopar · 2024-06-24T17:22:36Z

Alright, resolved the conflicts and everything looks happy. I'll let CI run to confirm nothing got messed up (it didn't) and then merge.

Myoldmopar · 2024-06-24T21:39:58Z

Thanks @amirroth !

amirroth added 2 commits June 11, 2024 16:42

Initial (and maybe only) commit

8efac20

Add state parameter, move GetFluidPropertiesData to init_state

6719e49

amirroth added DoNotPublish Includes changes that shouldn't be reported in the changelog Refactoring Includes code changes that don't change the functionality of the program, just perform refactoring labels Jun 12, 2024

amirroth added this to the EnergyPlus 24.2 milestone Jun 12, 2024

amirroth requested a review from Myoldmopar June 12, 2024 14:43

amirroth self-assigned this Jun 12, 2024

Myoldmopar added 2 commits June 12, 2024 09:45

Merge remote-tracking branch 'origin/develop' into InitStateStub

152d73a

Clang format

391ba01

amirroth commented Jun 12, 2024

View reviewed changes

rraustad approved these changes Jun 12, 2024

View reviewed changes

amirroth added 3 commits June 12, 2024 15:51

Move init_state below OpenOutputFiles

72f3fa0

Merge remote-tracking branch 'origin/develop' into InitStateStub

a99fce3

Merge develop (or something), fix conflict (not sure what happened here)

6e377c1

amirroth assigned Myoldmopar Jun 13, 2024

Myoldmopar added 2 commits June 24, 2024 11:59

Pull develop, resolve conflicts

f34d5c5

Merge remote-tracking branch 'origin/develop' into InitStateStub

1c9210c

Myoldmopar merged commit 4d45705 into develop Jun 24, 2024
15 checks passed

Myoldmopar deleted the InitStateStub branch June 24, 2024 21:40

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding stubs for `init_state` #10560

Adding stubs for `init_state` #10560

amirroth commented Jun 12, 2024

amirroth left a comment

amirroth Jun 12, 2024

amirroth Jun 12, 2024

amirroth Jun 12, 2024

amirroth Jun 12, 2024

rraustad Jun 12, 2024

amirroth Jun 12, 2024

amirroth Jun 12, 2024

rraustad Jun 12, 2024

mjwitte Jun 12, 2024

rraustad Jun 12, 2024

amirroth Jun 12, 2024

amirroth Jun 12, 2024

rraustad Jun 12, 2024

rraustad Jun 12, 2024

amirroth Jun 12, 2024

rraustad Jun 13, 2024 •

edited

Loading

Myoldmopar commented Jun 12, 2024

rraustad left a comment •

edited

Loading

rraustad commented Jun 13, 2024

Myoldmopar commented Jun 24, 2024

Myoldmopar commented Jun 24, 2024

Adding stubs for init_state #10560

Adding stubs for init_state #10560

Conversation

amirroth commented Jun 12, 2024

amirroth left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rraustad Jun 13, 2024 • edited Loading

Choose a reason for hiding this comment

Myoldmopar commented Jun 12, 2024

rraustad left a comment • edited Loading

Choose a reason for hiding this comment

rraustad commented Jun 13, 2024

Myoldmopar commented Jun 24, 2024

Myoldmopar commented Jun 24, 2024

Adding stubs for `init_state` #10560

Adding stubs for `init_state` #10560

rraustad Jun 13, 2024 •

edited

Loading

rraustad left a comment •

edited

Loading