Refactor and optimize storeCarbs function to limit carb equivalents of FPU conversion to 1.0 grams

[dnzxy]

Contents

This pull request refactors the storeCarbs function to improve readability, maintainability, and performance. The new version introduces a helper function processFPU to handle fat and protein units (FPUs) and ensure that each carb equivalent is at least 1.0 grams. The changes also include updated docstrings for better documentation.

Solves an issue where oref0's meal.js function ignores carb entries that are less than 1.0 grams.

Changes

• Extracted the FPU processing logic into a new helper function processFPU, which ensures that each carb equivalent is at least 1.0 grams by adjusting the interval if necessary.
• Added logic in new helper functionprocessFPU to adjust the interval when the calculated carb equivalent is less than 1.0 grams.
• Adjusted spread over time by computing duration again so total carb equivalents remain consistent
• Refactored the main storeCarbs function to be more concise and easier to understand.
• Introduced detailed docstrings to explain the purpose and functionality of the storeCarbs, processFPU, and calculateComputedDuration functions.

Detailed Changes

• Main Function (storeCarbs):
  • Synchronized processing using processQueue.sync.
  • Guarded against empty entries.
  • Delegated FPU processing to processFPU.
  • Stored normal carb entries and managed CoreData operations.
  • Notified observers after updating carbohydrate data.
• Helper Function (processFPU):
  • Processed fat and protein to calculate carb equivalents.
  • Ensured each carb equivalent is at least 1.0 grams by adjusting the interval.
  • Created future carb entries based on the adjusted interval.
  • Returned the future carb array and total carb equivalents.
• Duration Calculation (calculateComputedDuration):
  • Calculated the duration for processing FPUs based on predefined rules and the time cap.

[AndreasStokholm]

Looks great! Only had a single comment. Approved!

[MikePlante1]

LGTM.

Tested in a simulator with different settings and different fat/protein entries and looks to work as expected with carb entries converted from fat/protein entries never being less than 1 and reducing the number of entries to accomplish that. Any entry that converts to <1g carbs is rounded up to 1g. That sounds reasonable to me.

I compared the new code to the old code and didn't notice any discrepancies.

[aug0211]

Any entry that converts to <1g carbs is rounded up to 1g.

Is this the intended outcome, @dnzxy? Is rounding up part of the solution, or is the interval being extended so that rounding up does not need to take place? Or maybe both in some case?

@MikePlante1 can you share the test scenario you ran (FPU settings and example meal entry) you used to trigger rounding up?

[dnzxy]

Is this the intended outcome, @dnzxy? Is rounding up part of the solution, or is the interval being extended so that rounding up does not need to take place? Or maybe both in some case?

It is by implementation, so semi-intentional. We can either drop the "last" entry that needs rounding or we keep it, and round up to get to 1 gram. This stems from re-assigning equivalents to be 1g and still making sure that it "fits" into the total of equivalents.

I ran this example when I tested this feature during implementation with these settings:

• Two entries with fat 10g, protein 10g.
• FPU settings are set to what you can see in screenshot.
• First entry with old logic creates 17 entries at 0.4g, total of 6.8g
• Second entry with new logic creates 7 entries at 1.0g, total of 7g
• Side note: rebuild between entries

So the 7th entry should be 0.8g but is rounded up to be 1.0g, so +0.2g. Worst case would be an additional 0.9g of carbs some time down the line being added to COB and the algorithm would make a new dosing decision then.

I figured that is okay. What do you think?

[avouspierre]

tested with simulator and checked the code. LGTM !

[bjornoleh]

Any entry that converts to <1g carbs is rounded up to 1g.

Is this the intended outcome, @dnzxy? Is rounding up part of the solution, or is the interval being extended so that rounding up does not need to take place? Or maybe both in some case?

@MikePlante1 can you share the test scenario you ran (FPU settings and example meal entry) you used to trigger rounding up?

The rounding only applies to the last entry and will only amount to a single sub 1g rounding amount (if I understand correctly). I’d say this is acceptable.

You can test with the minimum amount of 1 g fat or 1 g protein, it will produce a single 1 g carb equivalent.

In the old implementation, the minimum values were 2 g fat or 5 G protein, which would produce 9-10 0,1 g equivalents (which are ignored). Less than 2 g fat or 5 g protein would result in nothing, which is more or less confusing too. I’d say it’s more intuitive to at least see that something is registered, and single gram changes are not expected to have much impact anyway.

[aug0211]

I agree, I wouldn't block merging for that rounding. It's better than what existed before!

The scenario that stands out the most is of course an entry of 0/1/0 or 0/0/1 (C/F/P), where we end up with 1g of "FPU carbs" being logged for Trio to consider in calculation.

As others said here, 1g extra is probably not a big deal, though I do wonder if it'd be even better to check around line 170 of CarbStorage.swift and apply rounding to the last entry, so that if it nets out to < 0.5g it is cancelled as 0g, and if it nets out to >= 0.5g, it rounds to 1g?

[MikePlante1]

If someone enters 1g of fat or protein, and expected it to log 0g of carb equivalents and therefore do nothing, they just shouldn’t have entered any fat or protein in the first place. Or is there a usecase for this that I’m not thinking of?

[aug0211]

Yes, of course - mathematically consider any numeral that results in a clean calculation into FPUs (based on the static formula and user defined settings), with 1g of fat or protein remaining as a leftover.

[bjornoleh]

I'll go ahead and merge this now with four approving reviews. Well done! 🚀