Re: Path to LilyPond 2.18

[Mike Solomon]

On 6 juin 2013, at 06:52, Keith OHara <address@hidden> wrote:

> Phil Holmes <mail <at> philholmes.net> writes:
> 
>> I believe we need to get fixes for 3363 and 3386 into the code base before 
>> cutting a release candidate. 
> 
> Issue 3363, the scripts on cross-staff things being grossly misplaced, 
> looks simple, but requires figuring out quite a lot of the sequencing
> of steps in layout.
> 
> Issue 3359, the infinite-recursion crash with the failed cross-staff tie,
> also worries me.  It shows that one of the properties with 'pure' in the
> name, 'pure-relevant-grobs' includes objects from different staves.
> Usually 'pure' things deal with spacing before line-breaking, but objects
> in different staves do not have their relative spacing known until later,
> after page-layout.  Having objects from different staves in a data set
> likely to be used for spacing before line breaking, seems like a mistake.
> 
>> I think 3386 is a simple revert, but don't 
>> have time to do it right now.
> 
> We probably cannot revert *all* of the patch that caused issue 3386 with
> the start braces, because that patch did solve issue 3242. I am pretty
> sure that my suggestion on your rietveld review will work.
> 

Hey all,

Sorry for being away for so long.

I have a 4-hour window today starting at 1:30 GMT I can spend fixing stuff.  
Could someone please send me minimal examples of things going wrong? Looking at 
the tracker, I don't have a good sense yet of what works, what doesn't, what's 
been reverted etc..

In general, my old lilypond war wounds start to itch when I look at functions 
like Beam::consider_auto_knee.  Anything that results in calling set_property 
or set_object is a recipe for trouble - I'm guessing that's partially in play 
here.

Grobs from different vertical axis groups being in a pure properties array 
shouldn't, in theory, cause a problem, as at that point the Y-common is a 
vertical alignment and one would just use pure minimum translations instead of 
minimum translations.  So the problem may be that the actual minimum 
translations are being calculated too early.

Cheers,
MS