Re: Suggestion to make sharps and flats persistent

[David Kastrup]

Kieren MacMillan <address@hidden> writes:

> Hi David,
>
>> LilyPond's input language has no representation for c-natural
>> as opposed to c-unkeyed-yet. Any kind of implementation would be doomed 
>> without
>> that, anyway.  Once you have that, it really becomes a tricky question
>> of where c-unkeyed-yet would get its final pitch.  And how this is
>> supposed to behave with regard to transposition: transposing half a step
>> up and down again should be a do-nothing, so you'd actually also need a
>> cis-unkeyed-yet .  Do we have a headache yet?
>
> \relative gives me a headache.
>
> My point is how is this really different, for those who want to give
> themselves headaches?
> Your answer doesn’t convince me it’s all that different — only unadvisable.

Before you get to implement something, you need to figure out its
semantics.  Otherwise you end up fixing stuff back and forth directed by
what feels "right" and "wrong" when actually there is no consistent rule
that could distinguish right from wrong and do only what is right.

-- 
David Kastrup