Re: Reduces algorithm time by prefinding footnoted grobs (issue4213042)

[Mike Solomon]

On Mar 1, 2011, at 10:19 PM, address@hidden wrote:

> Hi Mike,
> 
> I have a quick comment for you below.
> 
> Thanks,
> Patrick
> 
> 
> http://codereview.appspot.com/4213042/diff/24035/scm/define-stencil-commands.scm
> File scm/define-stencil-commands.scm (right):
> 
> http://codereview.appspot.com/4213042/diff/24035/scm/define-stencil-commands.scm#newcode37
> scm/define-stencil-commands.scm:37: footnote
> Shouldn't this be in `ly:all-output-backend-commands' instead?
> 
> (Note the docstring for `ly:all-stencil-commands' versus the one for
> `ly:all-output-backend-commands'.)
> 
> Some extra processing is done for elements of this list (see
> scm/backend-library.scm).  If the stencil expression is not defined in
> the backends, then it should be in the ly:all-output-backend-commands
> list instead to avoid unnecessary (though minimal) overhead.
> 
> 

Done.

I also was able to fix the problem Neil pointed out before, although it 
required certain things once marked const to be downgraded to not const in 
certain places.  I tried to make this minimal & re-constify their constness 
ASAP.  The only place where this may be problematic is 
Page_breaking::space_systems_on_1_page, where the vector of lines floats around 
not-const for the entire function.

Cheers,
MS