On Sat 21 Apr 2018 at 23:33:44 (-0400), Kieren MacMillan wrote: > I'm running Frescobaldi 2.20.0 on Mac OS X 10.12.6. > > 1. I open the Lilypond file(s) in Frescobaldi, and once the first PDF is > generated, I open that in an external PDF viewer (usually Skim; sometimes > Preview), because I find Frescobaldi’s built-in viewer to be of lesser > quality. > > 2. As I recompile, the external application shows the changes (on my second > monitor, which is very nice to have!). If I work long enough on a file, it > eventually happens that the external application stops updating on > compilation. At that point, I can look in the log to find the temp path that > Frescobaldi is using, e.g., > > Layout output to > `/var/folders/l3/8qtf8wts5kj93q4w13f4qdfc0000gn/T//lilypond-4TMu7P' > > and open the newly-compiled PDF in that folder. Beyond that point, no amount > of recompiling "jump-starts" the non-temp version of the PDF. > > 3. Quitting and restarting Frescobaldi always "jump-starts" the non-temp > version of the PDF — but after working on the file a certain > [indeterminate!?] period of time, the problem appears again.
I'm not a Mac user, so this is just a guess based on zero evidence. On linux, I'm used to running LP in one xterm and viewing the PDF with xpdf in mc in another one. I have to press r in the PDF to refresh it (though changing page, or resizing the PDF, will have the same refreshing effect). If I press r at the moment that LP is converting PS→PDF, the effect can be bizarre (eg staves, stems and beams but nothing else) because xpdf is reading a file that LP is still writing. IOW there's no file-locking to protect the processes from each other. Now imagine that there IS file-locking. LP would lock the PDF file each time it revised it. If I pressed r at that time, I expect xpdf might silently ignore the request and continue to display the old contents. But what if xpdf locked the PDF file each time I pressed r until it had read the file—what would LP do if it hit a lock? One reaction might be to write to a nonce file instead, so that it can return control to the user (and update its graphics panel). Just an idea; no brickbats please. I have no idea whether Macs use file-locking here, and I don't use Frescobaldi either. Cheers, David. _______________________________________________ lilypond-user mailing list lilypond-user@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-user