Re: [Lilypond-auto] Issue 2366 in lilypond: THANKS needs updating or deleting

[lilypond]

Comment #18 on issue 2366 by address@hidden: THANKS needs updating or deleting
http://code.google.com/p/lilypond/issues/detail?id=2366

Regarding comment #16: "irrelevant divergence" does not make sense.  You  
likely mean either "unnecessary" or "temporary".  "Surely it would be  
easier if master was (as much as possible) a superset of stable/2.16": of  
course, and we have given this easiness a year of release candidates to  
show its worth.

The reality is that we won't get a stable release without sacrificing about  
two of four from automated decision-finding, conscientious release  
criteria, quality of release, and coherence of master with stable.

There is not just the problem of a failing build after structural changes:  
there also are problems of incomplete tarballs, dead links in documentation  
and other stuff.

Yes, every divergence of master from stable is a nuisance, and it would  
be "easier" to forego all of them, since no change to master happens in the  
expectation of creating problems rather than solutions.

Now we only need to placate the conscience of a single person rather than  
that of an automated process.  That's a single safety net rather than a  
complete system, and while I am responsible for it, I won't lightly remove  
that as well.

What is the cost? A commit reverted a short time after 2.16.0 and 2.17.0,  
and a state of 2.16.0 diverging in one respect from the 2.16.1 situation  
that is supposed to supplant it about a month later and that already  
clearly is desirable because of other reasons.

Comment #17 does not convince me that I would show good engineering sense  
by admitting a more invasive change of 2.16 right now, untested in any  
unstable release.  In particular when I have already rejected the inclusion  
of build system surgery intended to decrease the impact of oversights in  
this kind of patch.