Re: [Denemo-devel] Chord symbols / dead lilypond processes

[Richard Shann]

On Sun, 2013-05-05 at 11:54 +0200, Andreas Schneider wrote:
> Am 05.05.2013 11:44, schrieb Richard Shann:
> > I see the problem - the call to lilypond to get the installed lilypond
> > version is using g_spawn_async but not waiting for the process, it is
> > just working by accident and leaving the zombie. I have disabled this as
> > we are moving to packaging lilypond with denemo anyway. But we are
> > currently hardwiring the version in print.c where it should be given by
> > the gub build system.
> > Fixed in git.
> 
> In my opinion Denemo should work with a system installation of lilypond.
It does by default on Unix, on windows by default it uses the one
packaged with Denemo. You can choose a particular version via the
preferences.
> Maybe a user wants to use a new feature of the development version of
> lilypond, or I can imagine several other reasons. Why don't you get the
> system lilypond version on startup of Denemo, using "lilypond
> --version", and store it in a variable?
yes, that is what should have been done. Instead on Unix it was running
lilypond --version each time, and doing it with faulty code.
What it is doing now is a reasonable starting point - Denemo is putting
a version number into its LilyPond text that is a literal value in the
Denemo source code. This is a version number that covers all the syntax
used in current scripted commands.
People may have written scripts since then expecting later syntax and if
they run later versions of LilyPond via Denemo it will work, albeit with
a warning from LilyPond that the file version is not up to date.
Avoiding these warnings was the purpose of getting the actual LilyPond
version but, as I say, it wasn't being done correctly, so I think we are
better putting up with the warnings.

Richard