| description | Index of Sendra design decisions: unasserted statuses under test, how exit codes are split, script sandboxing and the precedence chain. |
|---|
The rationale behind the design choices in Sendra's schema and CLI that could reasonably have gone another way - an ADR-style collection, one file per decision. For the shape of a feature (what a field is called, what it does), see docs/reference.md; these are the "why", not the "what".
- Why
testignores an unasserted status: why a404nobody wrote an assertion against does not fail asendra testrun, and why apost_requestscript counts as a check the same way an assertion does. - Exit codes: why the numbers are split the way they are:
why
4is a separate number from3, why an unsendable request undertestexits1rather than4, and how these codes rank against each other for a whole collection. - Sandboxing: the guarantees behind
pre_request/post_request: no filesystem, no network, nothing registered into the script interpreter beyond the request/response themselves. - Precedence, start to finish: every layer that can
decide a value for a request (defaults, config, environment, captures, CLI
overrides,
pre_request), stated as one chain.