Trevor Daniels wrote Saturday, August 18, 2012 10:03 PM
> I don't know what Phil thinks was fixed between 2.15.18 and 2.15.21, > but I do suspect the symptoms of the problem or problems changed > from release to release, and may even depend on the Windows setup. I've just turned up an email from me to Carl dated 4 Jan 2012 which suggests the problem /did/ change from the originally reported msvcrt one to this in 2.15.23: > File "c:\Program Files\LilyPond\usr\lib\python2.4\subprocess.py", line > 549, in __init__ > self.stdout = os.fdopen(c2pread, 'rU', bufsize) > OSError: [Errno 22] Invalid argument > Lilypond-book returned code 1 Then we found problems with subprocess.Popen and subprocess.PIPE, with sleep (which was a Solaris hack), and with the setting for universal_newlines. And the definition of cmd was wrong for Windows. All these were fixed by Graham's patch, which was actually pushed to git on 15 Jan. Trevor _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
