This repository was archived by the owner on Sep 10, 2025. It is now read-only.

Description
With #1599 most tests have utf-8 roundtrip tests with the exception of the IWSLTs. It seems like the xml.etree.ElementTree.fromstring has no mechanism for recovering from error-inducing unicode (which is likely in randomly-generated strings). lxml has such a mechanism, but I'd hate to add a userland dependency for the sake of testing.
This issue records the TODO and acts as a place to brainstorm alternative mechanisms for testing roundtrip utf-8 encode/decodes in IWSLT.