Re: recompilation failure, strange

[David Kastrup]

胡海鹏 - Hu Haipeng <address@hidden> writes:

> Hello,
>   I modified some notes in my overture, and tried to recompile it. But
> Lilypond 2.15.20 suddenly aborted without any messages left in the
> log. It's at about bar 160, but I can't detect any error, because it
> compiled well when i use 2.13. I don't use anything changed during
> these months, except markuplines. I know the files are big, but could
> anyone tell me why lilypond aborted this way and how to find the
> error?

First, of course you use something changed during this month, and

convert-ly -ed *.ly

consequently makes a lot of changes.  It can't help, however, with
spurious newlines created by editing mistakes or botched copy&paste
jobs.

So I recommend that you first try to get your document into a shape
where it would actually compile with a version of 2.13, then try
convert-ly on it.

And then there is one thing that is quite non-obvious and needs fixing:
a comment.

    % direction (either #UP or #DOWN)

# and $ in comments inside of #{ ... #} are not completely
unproblematic.  $ was much more dangerous, now both $ and # are a bit
more low-profile.  I doubt that except for quite a short time in the
life cycle of 2.15 that this has ever worked.

With current 2.15, it will start working again if you put a space after
#DOWN.

The thing is that the Scheme reader engages after # and reads one
expression, and as opposed to a space, it can't silently swallow ) and
it can't back it up either.

I don't really think that creating convert-ly rules for this case
(defusing comments inside of #{ ... #}) makes all that much sense, but
it would be possible.

-- 
David Kastrup