Re: Failed compiling a single lsr-snippet, which does not fail whenrunning the whole lsr

[David Kastrup]

"address@hidden" <address@hidden> writes:

> On Apr 3, 2012, at 1:10 PM, Phil Holmes wrote:
>
>> 
>> We need to check what the script is doing and how it responds to
>> failing snippets like this.  I'm not in a position to do this right
>> now, but let's not forget it's there.  I've put it on the tracker as
>> http://code.google.com/p/lilypond/issues/detail?id=2466
>> 
>
> It's also important to establish if this is a critical regression or
> not.  The snippet dug into the bowels of LilyPond and used lots of
> functionality that's not put into play often.  So its failure to
> compile may be pointing to an unexposed regression.  Perhaps it'd be
> wise to label this as critical now and then downgrade it once we
> confirm that the compilation failure isn't coming from anything
> directly involved in running the LilyPond executable.

There are things like bleedover.  See
<URL:http://code.google.com/p/lilypond/issues/detail?id=2449> for one
example where a "user interface" was added that results in changes
affecting the entire rest of a multi-file session.  If a later snippet
relies on a define-event-class in an earlier snippet, bleedover will
make this reach into further sessions.

git grep define-public Documentation/snippets

returns several snippets that _export_ definitions (I have not checked
whether this means that they will spill over, but I would not be
surprised if it did).

Any snippet that reaches into the bowels of LilyPond causing a
_permanent_ change there surviving beyond the session is a mistake.

It is possible that here one snippet depends on another snippet.

What are the error messages in this particular case?

-- 
David Kastrup