[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Thu, 06 Sep 2012 00:15:45 +0200
Mozilla/5.0 (X11; Linux x86_64; rv:15.0) Gecko/20120827 Thunderbird/15.0
On 2012-09-05 18:15, David Kastrup wrote:
Karim Haddad <address@hidden> writes:
Apparently, in the new stable version 2.16, the "\override Stem #'flag
= #modern-straight-flag" is broken. , i.e this has no effect, the
flags stay traditional. or maybe i missed something.
Anyhow, Thank you a lot to all developers who contributed and to all of you.
Maybe we should hammer home the existence of convert-ly right in our
stable release announcements. Its discoverability is likely not
sufficient for making people pick the smoothest upgrade path.
This is a particularly nasty example, since lilypond seems to work fine
without any warning or error. Some functionality simply silently stopped
working and suddenly produces a different output without any indication!
If we really need to change the lilypond language  (and I also count
changes to overrides as an relevant change in the lilypond language),
then at least we should make sure that either the old syntax keeps
working (via a compatibility layer or by keeping some old definitions),
or make sure that lilypond FAILS with the old syntax (or at least give a
warning!). Giving the user the impression that everything worked, even
though things have changed, is the worst case and a good way to make
professionals loose confidence in lilypond.
And you can't really require all users to run convert-ly on any file for
any small lilypond update (with the package systems in the various Linux
distributions, the users might not even be aware when a new minor
version of lilypond was installed!)
I have been badly burnt such a situation a while ago. Just before
sending one of my scores to the print office (thank god it was before,
not after!), I noticed that lilypond had silently discarded most of the
text crescendos! The compilation didn't print out any warning, it simply
didn't print "cresc." in certain cases, because the way text crescendo
were implemented had changed meanwhile.
 Note, however, that ANY change, even a very small, subtle change, is
a really grave argument for a music publisher against using lilypond.
I wrote a huge piece (~95 pages full score, 23 orchestra instruments,
choir, etc) a few years ago. I didn't count the hours it took me last
month to bring it up to date with the latest lilypond version (I had to
visually compare the whole full score and all instruments to make sure
that nothing had been lost).
At that time, I really, really, really cursed lilypond and its frequent
Reinhold Kainhofer, address@hidden, http://www.kainhofer.com
* Financial & Actuarial Math., Vienna Univ. of Technology, Austria
* http://www.fam.tuwien.ac.at/, DVR: 0005886
* Edition Kainhofer, Music Publisher, http://www.edition-kainhofer.com