Hello,
On 10 August 2013 08:23, Keith OHara <[email protected]> wrote: > On Fri, 09 Aug 2013 23:58:25 -0700, James <[email protected]> wrote: > > [...] > >> Then I got this: >> >> jlowe@jlowe26vm ~/Desktop$ lilypond -dpreview -dresolution=150 -o hello >> orchestra.ly >> GNU LilyPond 2.17.19 >> > > programming error: Parsed object should be dead: #<Context_mod ((push >> VerticalAxisGroup #t remove-empty) (description Remove staves which are >> considered to be empty according >> to the list of interfaces set by @code{keepAliveInterfaces}.))> >> continuing, cross fingers >> > > Success: compilation successfully completed >> >> >> So whatever is going on is obviously not consistent :) >> > > Is it significant that I seem to get the additional programming error when >> I change the directory of where I -o the output? >> >> > This "should be dead" error comes up randomly but rarely, and only on > builds with debugging enabled. > Looking through your results, it seems that it could be two independent > issues: the inconsistent "should be dead" error, the reliable "cyclic > dependency" messages ... but none of the crashes that Phil and Petr see. > > But, don't you have one of the systems that had trouble with the > orchestra.ly file during a `make doc`? Or was your system giving an > unexplained error on `make doc` only while testing issue 3255? > > Yes I do '-disable-optimising' when I built the binary - that is I assume, with debugging enabled - the 3255 was with 'chords-funky-ignatzek', never had a problem with orchestra on that. Orchestra.ly was one of DK's patches or rather I spotted it while compiling one his patches, but it 'went away' and I cannot recall if it was a make doc or not. I always run patchy test with make doc as well (in case you wondered). James
_______________________________________________ lilypond-devel mailing list [email protected] https://lists.gnu.org/mailman/listinfo/lilypond-devel
