Looks for prebroken pieces of dead items in the pure relevant function.

Looks for prebroken pieces of dead items in the pure relevant function.

Even if the item is dead, its prebroken pieces may not be.  We need to check
them and build them into the pure skylines of axis groups used by the
align interface

[MikeSol, 2013-10-28 08:48:44]

Adds comment.

[Keith, 2013-10-30 07:32:18]

This looks correct. 

It includes a pointer to the potential Clef in every single measure in
'pure-relevant-grobs', but the user of that list checks whether the Clef would
be visible.

The output with annotate-spacing=##t looks correct.  The bar numbers are still
tentatively placed against the top staff-line, but now that placement puts them
only 0.5 staff-space above the upper limit of the system extent.

Adding Clefs to the space reservation can the system height because of the
system skylines during the line-breaking stage are two potentially-misaligned
boxes: begin_of_line_extent and rest_of_line_extent.  These boxes align at the
top of the system, so the staff-lines are conceptually discontinuous.  In the
example in the issue, while Clefs were ignored, the bar number sat fully above
the system, taking 2.25 staff-spaces additional height rather than the proper
0.5

If anyone put lots of treble-clef staves in a system, and decreased the
minimum-distance between staves, he will see over-full pages before this patch.

I do not understand the Todo: comment; nothing should have been suicided at this
stage.  What is the life cycle of a Clef and the copies made for line-breaking
possibilities?  Is the never-printed original Clef /ever/ is_live(), or just its
broken copies?

https://codereview.appspot.com/18090043/diff/20001/lily/axis-group-interface.cc
File lily/axis-group-interface.cc (right):

https://codereview.appspot.com/18090043/diff/20001/lily/axis-group-interface....
lily/axis-group-interface.cc:227: extract_grob_set (me, "pure-relevant-grobs",
elts);
The use of pure-relevant grobs here...

https://codereview.appspot.com/18090043/diff/20001/lily/axis-group-interface....
lily/axis-group-interface.cc:283: if (g->pure_is_visible (start,
visibility_end))
tests each Clef to see if it would be printed on the tentative line.

[mike7, 2013-10-30 08:09:53]

On Oct 30, 2013, at 8:32 AM, k-ohara5a5a@oco.net wrote:

> I do not understand the Todo: comment; nothing should have been suicided
> at this stage.  What is the life cycle of a Clef and the copies made for
> line-breaking possibilities?

The original clef is suicided in the handle_prebroken_dependencies function,
which is the same place that the new begin and end of line grobs are created
(see break-substitution.cc).

> Is the never-printed original Clef /ever/
> is_live(), or just its broken copies?
> 

It is live before handle_prebroken_copies.  In this function, the broken copies
are created, and then it is suicided.

Conceptually, I don’t like the idea that a suicided grob contains useful
information.  You’ll see in the suicide function that its entire cache is
cleared out (Grob::suicide) save one or two things (Item::suicide).  I think
everything should be cleared out.  Otherwise, the code is less maintainable -
developers will have to remember what type of information suicided grobs hold
and what type they don’t.  That’s why I created
https://code.google.com/p/lilypond/issues/detail?id=3636.

Cheers,
MS

[Keith, 2013-11-01 06:49:19]

On 2013/10/30 08:09:53, mike7 wrote:
> On Oct 30, 2013, at 8:32 AM, mailto:k-ohara5a5a@oco.net wrote:
> 
> > I do not understand the Todo: comment; nothing should have been suicided
> > at this stage.  
> 
> The original clef is suicided in the handle_prebroken_dependencies function,
> which is the same place that the new begin and end of line grobs are created
> (see break-substitution.cc).
 
> Conceptually, I don’t like the idea that a suicided grob contains useful
> information.  

I see now.
It seems that !is_live() means that we are sure the Grob will never be printed,
yet we need to keep the data structure around for some other reason.  You might
put your name on the "Todo" comment so people know the point of view is from
somebody other that the original author of the handle_prebroken_dependencies()
function.

The patch looks good.

[mike7, 2013-11-01 07:15:29]

On Nov 1, 2013, at 7:49 AM, k-ohara5a5a@oco.net wrote:

> On 2013/10/30 08:09:53, mike7 wrote:
>> On Oct 30, 2013, at 8:32 AM, mailto:k-ohara5a5a@oco.net wrote:
> 
>> > I do not understand the Todo: comment; nothing should have been
> suicided
>> > at this stage.
> 
>> The original clef is suicided in the handle_prebroken_dependencies
> function,
>> which is the same place that the new begin and end of line grobs are
> created
>> (see break-substitution.cc).
> 
>> Conceptually, I don’t like the idea that a suicided grob contains
> useful
>> information.
> 
> I see now.
> It seems that !is_live() means that we are sure the Grob will never be
> printed, yet we need to keep the data structure around for some other
> reason.  You might put your name on the "Todo" comment so people know
> the point of view is from somebody other that the original author of the
> handle_prebroken_dependencies() function.
> 

Ok.

Cheers,
MS

[MikeSol, 2013-11-01 07:18:53]

Adds my name to TODO

[MikeSol, 2013-11-01 07:25:50]

Adds TODO

[lemzwerg, 2013-11-01 07:31:04]

https://codereview.appspot.com/18090043/diff/60001/lily/axis-group-interface.cc
File lily/axis-group-interface.cc (right):

https://codereview.appspot.com/18090043/diff/60001/lily/axis-group-interface....
lily/axis-group-interface.cc:507: should probably not check for suidided items
or NULL pointers
s/suidided/suicided/

[MikeSol, 2013-11-01 07:34:19]

Better spelling

[mike7, 2013-11-01 07:34:55]

On Nov 1, 2013, at 8:31 AM, lemzwerg@googlemail.com wrote:

> 
>
https://codereview.appspot.com/18090043/diff/60001/lily/axis-group-interface.cc
> File lily/axis-group-interface.cc (right):
> 
>
https://codereview.appspot.com/18090043/diff/60001/lily/axis-group-interface....
> lily/axis-group-interface.cc:507: should probably not check for suidided
> items or NULL pointers
> s/suidided/suicided/

Done.

Cheers,
MS