Re: [Lilypond-auto] Issue 2449 in lilypond: Redesign stream event class representation

[lilypond]

Updates:
        Labels: -Patch-countdown Patch-new

Comment #21 on issue 2449 by address@hidden: Redesign stream event class  
representation
http://code.google.com/p/lilypond/issues/detail?id=2449#c21

Further deglobalize the event class hierarchy (part of issue 2449).

This is more a discussion basis than anything else: the most important
change is likely making the "class" field in stream events contain the
complete class (a list of event types) rather than just the basic
event type symbol.  The reason behind that is that the event class
hierarchy is more or less a function of the current global context,
and dispatchers are not really tied into contexts, so the respective
context-relevant information is not easily accessible.  So the event
class hierarchy gets consulted when _creating_ the stream event.

At the presented state of the code, the event class hierarchy _is_
still global.  This is more or less juggling code around.

One consequence is that it is becoming less realistic to interpret
stream events written to file.  After all, the set of event types is
intended to become open-ended, and the reader of a file might not know
about new types.

Anyway, the current state are about three commits:


Make detection of listened events a property of dispatchers instead of  
global


Define and use event-class-cons


Change class field in stream events to contain the complete event type list

http://codereview.appspot.com/6121050