Re: (scheme-)engraver in 2.20/21

[Dan Eble]

On Sep 24, 2020, at 07:52, Aaron Hill <lilypond@hillvisions.com> wrote:
> 
> On 2020-09-24 4:42 am, Dan Eble wrote:
...
>> This change was intended.
>> https://gitlab.com/lilypond/lilypond/-/merge_requests/292
>> specifically, commit fe9242659305dce587bd1fcdcc7b0ac62df25ad6
> 
> Does this not actually increase complexity?  An engraver that only needed to 
> handle start-translation-timestep now needs to handle initialize as well or 
> else it misses the opportunity to set up state.

Restate "An engraver that only needed to handle …" as "an engraver that would 
only work when created at the start of a piece, and would fail to initialize 
itself properly in a new context created in mid piece" and I think you will 
understand what's missing from your accounting.
—
Dan