Bug: Inconsistent behavior for missing values in coordinate_compression.py

[tommasobaiocchi]

Repository commit

a0b0f41

Python version (python --version)

Python 3.10.7

Dependencies version (pip freeze)

Not applicable - this is a logical bug, not dependency-related

Expected behavior

Consistent behavior across all methods when handling missing values:

• Either ALL methods should raise exceptions for invalid inputs
• Or ALL methods should return consistent sentinel values
• The behavior should be clearly documented
• No silent failures mixed with exceptions

Actual behavior

Inconsistent behavior:

1. compress(80) returns -1 (silent failure)
2. coordinate_map[80] raises KeyError (exception)
3. decompress(5) returns -1 (silent failure)

This mixed behavior can cause hard-to-debug issues.

[DevPatel-11]

Hello @tommasobaiocchi ,
Can I work on this issue ?

[AyushYadav256]

please assign me this issue

[tommasobaiocchi]

Hi @DevPatel-11 and @AyushYadav256!
Thanks for your interest in fixing this issue!

I agree that @DevPatel-11 should have priority since they commented first.

@DevPatel-11 - please coordinate with the maintainers on the preferred approach (exceptions vs sentinel values) before implementing.

@AyushYadav256 - there might be other similar issues in the repository that could use your help!

Looking forward to the fix!

[DevPatel-11]

Hi @tommasobaiocchi ,
The compress(), coordinate_map[] and decompress() functions you are talking about are from the data_compression\coordinate_compression.py, right ?

[tommasobaiocchi]

Yes @DevPatel-11, exactly!

[DevPatel-11]

Hello @tommasobaiocchi ,
I read the code. I think that the solution to this is instead of using coordinate_map[80] we can simply create a get_compressed(val) function.

def get_compressed(self, original: int) -> int:
    return self.coordinate_map.get(original, -1)

Otherwise, we directly use coordinate_map.get(80, -1).

[tommasobaiocchi]

@DevPatel-11 Thanks for the concrete proposal!

You're right that using .get(80, -1) would make the behavior consistent, but I'm concerned about a few architectural issues:

Current Inconsistency:

• compress(80) → -1 (silent failure)
• coordinate_map[80] → KeyError (exception)
• decompress(5) → -1 (silent failure)

Problems with -1 as Sentinel:

1. Ambiguity: -1 could be a valid compressed value in some datasets
2. Error Masking: Silent failures make debugging harder
3. API Design: Public methods vs internal dict access should be consistent

Better Approaches:

Option A:

def compress(self, original):
    if original not in self.coordinate_map:
        raise ValueError(f"Value {original} not in original array")
    return self.coordinate_map[original]

Option B:

_MISSING = object()
def compress(self, original):
    result = self.coordinate_map.get(original, _MISSING)
    return result if result is not _MISSING else None

What do you think about these more principled approaches?

[Amitverma0509]

Option A: Explicit Exception
Pros:
1.Clear contract: compressing a non-existent value always fails loudly.
2.No ambiguity: -1 or any other value can safely exist in the mapping.
3.Easy to debug: errors are immediate and informative.

Cons:
1.Slightly less forgiving—caller must handle exceptions if uncertain of the value.
2.Could interrupt batch operations unless wrapped in try/except

Option B: Sentinel / None
Pros:
1.Safe: never raises an exception.
2.Flexible: caller can check for None and handle missing values gracefully.
3.Avoids magic numbers like -1.

Cons:
1.API may silently return None if the caller forgets to check.
2.Slightly more complex to implement if _MISSING sentinel is needed in multiple places.

My Recommendation
1.If API is meant for internal use or you expect callers to always know valid values → Option A.

2.If API is user-facing and missing values are expected/normal → Option B

[tommasobaiocchi]

@Amitverma0509 Excellent analysis.

[DevPatel-11]

Hey @tommasobaiocchi , should I mention cclaus to discuss the solution ??
I am new to the repository. I don't know any maintainers.

[tommasobaiocchi]

@DevPatel-11 Let's wait for maintainer feedback on the preferred approach.
They usually monitor new issues and will provide guidance when available.

[Aditya-Singh27]

Hi, I’d like to work on this issue. Can I take it?