Re: translation-type? vs. backend-type?

[David Kastrup]

Lukas-Fabian Moser <lfm@gmx.de> writes:

> Hi David,
>>> Ok, so then I remain unclear as to why the original code seemingly
>>> worked with 2.12. But at the end of the day, I don't care that much:
>>> 'translation-type it is then, for what should be a \set'table context
>>> property.
>> I repeat:
>>
>>>> the change flagging the preexisting problem may possibly be
>>>>
>>>> commit e44bc2cc48f9fb714603bdb5f8db532d55374087
>>>> Author: Neil Puttock <n.puttock@gmail.com>
>>>> Date:   Thu Apr 30 20:29:09 2009 +0100
>>>>
>>>>       Add type-checking to \tweak.
>>>>
>>>> in version 2.13.1
>
> Sorry for being thick - the use of the word \tweak confused me.

Well, there has been an increasing tendency of enforcing the type system
over the years, starting with warnings and ending with flat-out refusals
to assign bad values.  I have not looked at the actual commit itself
here but just looked for a few pickaxe strings (git log -S ...) and log
messages (git log --grep ...).  So it might or might not have been that
exact commit.

> I do not have the old development versions available, so my manual
> bisecting is very much on the coarse side. But I can confirm that:
>
> For the following (WRONG!) definition of a context property
>
> #(set-object-property! 'myprop 'backend-type? integer?)
> #(set-object-property! 'myprop 'backend-doc "test property")
>
> (A) LilyPond 2.12.3 complains about \set Staff.myprop = 42 but
> executes it (the property can be retrieved afterwards).
> (B) LilyPond 2.12.3 does NOT complain about
>   \context Staff {
>     \applyContext
>     #(lambda (ctx)
>        (ly:context-set-property! ctx 'myprop 84))
>   }
> (and executes it).
>
> LilyPond 2.14.0 still behaves like 2.12.3
>
> LilyPond 2.14.2 complains about (A) and (B), but executes the
> assignments nevertheless.
>
> LilyPond 2.16.2 still behaves like 2.14.2
>
> LilyPond 2.18.0 complains about (A) and (B) and refuses to execute the
> assignments.
>
> And of course, using a correct definition (using 'translation-type?),
> nothing changes between 2.12.3 and 2.22.0.

That would be roughly in line with my overall expectations.

-- 
David Kastrup