`read()` silently returns default values on all failure paths with no mechanism for downstream error detection

[GaneshPatil7517]

Problem Description

Every failure path in read() file not found, empty after max retries, parse error, ZMQ timeout returns the initstr_val default. There is no way for the calling control loop to distinguish between "received real data that happens to equal the default" and "failed to read, got fallback." In a control system, feeding stale/default values into a controller without the controller knowing they are stale can cause unsafe actuation.

Technical Analysis

Python (concore.py, read()):

# On FileNotFoundError:
ins = str(initstr_val)
s += ins  # <-- appends default to s, so unchanged() may see it as "new data"

# On max retries:
return default_return_val

# On parse error:
return default_return_val

Critically, the FileNotFoundError path appends the default to s, which means unchanged() will see this as a data change and break the spin-wait loop allowing the node to proceed with default data as if it were real input.

The C++ version has the same behavior:

catch (...) { ins = initstr; }   // fallback to initstr, no error flag
s += ins;                        // contributes to unchanged() check

Proposed Improvement

Return a status-bearing result type instead of raw values:

# Option: return (data, is_valid) tuple
def read(port, name, initstr):
    # ... on success:
    return (payload, True)
    # ... on failure:
    return (default, False)

Or provide a per-read error counter that nodes can check:

concore.last_read_status  # SUCCESS, TIMEOUT, PARSE_ERROR, FILE_NOT_FOUND

This is backward-compatible if callers that don't check the status simply ignore the second return value.

Impact

• Reliability: Controllers operate on stale data without knowing safety risk in neuromodulation
• Real-time: No distinction between "slow producer" and "dead producer"
• Julia impl: Opportunity to use Julia's Union{T, Nothing} or error types from the start
• Cross-language: All implementations share this gap; needs protocol-level error signaling

[GaneshPatil7517]

hey @pradeeban sir please can i work on this.....

[pradeeban]

@GaneshPatil7517 ok, go ahead.