On Mon, Aug 18, 2025 at 04:16:21PM +0200, Pablo Neira Ayuso wrote: > On Wed, Aug 13, 2025 at 07:05:35PM +0200, Phil Sutter wrote: > > This series consists of noise (patches 1-13 and most of patch 14) with a > > bit of signal in patch 14. This is because the relatively simple > > adjustment to JSON output requires minor adjustments to many stored JSON > > dumps in shell test suite and stored JSON output in py test suite. While > > doing this, I noticed some dups and stale entries in py test suite. To > > clean things up first, I ran tests/py/tools/test-sanitizer.sh, fixed the > > warnings and sorted the changes into fixes for the respective commits. > > Reviewed-by: Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx> Series applied, thanks! > I will follow up with a patch to partially revert the fib check change > for JSON too. Hmm. That one seems like a sensible change and not just a simplification of output. I guess if we take this approach seriously, we should agree on (and communicate) an upgrade path for JSON output. In detail (from the top of my head): 1) What changes are considered compatible (and which not) 2) In which situations are incompatible changes acceptable 3) How to inform users of the incompatible change I'd suggest something like: 1) Additions only, no changes of property values or names 2) Critical bug fixes or new (major?) versions 3) Bump JSON_SCHEMA_VERSION? Or is the "version" property in "metainfo" sufficient if bumped anyway? Cheers, Phil
