Hi everyone, As brought up in some meetings some of you might have attended, there are some issues right now with the SQL++ grammar being out of sync with the documentation and railroad diagrams. A similar situation exists with undocumented features that don't have APEs, that were introduced after we accepted the APE process.
It is already enforced with votes in Gerrit that when patches are accepted into master, they do not cause tests to regress and are reviewed by someone else. My proposal is that we treat these two issues the same way by having a new axis or repurposing one of the existing vote axes to disallow merging of a change in the grammar without updating the railroad diagrams and grammar docs, and as well as automatically flagging changes adding over 1000 lines to reference an APE in the commit message or specifically state that this is not a new feature. Thoughts? - Ian
