0x1F as nested namespace separator for REST URL paths conflicts with Servlet spec

[westse]

Feature Request / Improvement

The current Iceberg REST spec uses 0x1F to separate the parts of a multipart (nested) namespace parameter in URL paths and elsewhere. However this conflicts with the Servlet spec which calls for rejecting control characters (encoded or not), including the unit separator (0x1F), as "suspicious" and to return 400. Servlet containers enforce this (e.g. see jetty/jetty.project#11495), making implementing the Iceberg REST spec difficult.

Should a different, or at least alternative, separator character be selected to avoid conflicting with the Servlet spec?

Query engine

None

[arthurliu-db]

+1, we're running into a similar issue since we use Armeria web servers and web clients which also rejects requests with 0x1F in the URL path

[github-actions]

This issue has been automatically marked as stale because it has been open for 180 days with no activity. It will be closed in next 14 days if no further activity occurs. To permanently prevent this issue from being considered stale, add the label 'not-stale', but commenting on the issue is preferred when possible.

[westse]

not-stale

[ebyhr]

@nastra I think we can close this issue now. #10877 didn't automatically close this issue as the PR description contains a redundant #.