skip_digestion Behavior#

ArgDigest supports an optional bypass parameter to skip digestion at call time.

This mechanism is useful, but it should be treated as a controlled exception to normal behavior, not as the default usage pattern.

Typical setup#

@arg_digest(config="mylib._argdigest")
def get(molecular_system, selection=None, skip_digestion=False):
    ...

with:

# mylib/_argdigest.py
SKIP_PARAM = "skip_digestion"

What happens when bypass is enabled#

When bypass is enabled, argument digesters and pipeline rules are both skipped, so raw function arguments flow directly into your function logic.

Safe usage guidance#

Use bypass only for:

  • internal debug scenarios,

  • migration bridges,

  • controlled performance probes.

In most libraries, bypass should not be exposed as a default public behavior for end users, because it weakens the consistency guarantees of the API boundary.

Testing recommendation#

When bypass exists, test both paths:

  1. default path (digestion on),

  2. bypass path (digestion off).

This two-path testing is important: it prevents bypass behavior from drifting silently as digestion logic evolves.

Next#

Continue with Pipeline Design Patterns.

It bypasses both axes#

skip_digestion=True skips the function contract as well as the digesters. That is what makes it an escape hatch, and it does not make the call less safe than an undecorated function: with the contract skipped, an unexpected keyword reaches Python, which refuses it with its own TypeError.