I was having problems, getting MIDI-clock / realtime-MIDI to work in PureData. When I start receiving a MIDI-clock-signal freshly for the first time after starting PureData everything works as expected. PureData received the realtime MIDI data. (I use this patch for deriving the clock-data: http://little-scale.blogspot.de/2013/03/dealing-with-midi-clock-in-pd-sync-pd.html).

But now... when I stop the clock and restart the thing, I always got the error "warning: MIDI timing FIFO overflowed" in pd-console. I am using Ableton Live 9 as host - MIDI is going over the IAC driver on Mac OS-X. Search on the web only brought old websites and no solution to this, so I experimented a little bit myself and maybe found a solution:

What helped was the following: By coincidence, I went to the "preferences..." setting-panel. There I turned of the "Realtime Termination"-flag (I have the version in German, there is it called "Echtzeit Terminierung" - maybe the English translation differs.)

Does anyone knows what the Reatime-Termination Flag does? Any why this is solving the realtime-MIDI-issues that occur with receiving MIDI-clock signals? I just want to be sure, that this is really solving the problem.

+1. Ran into the same issue, but with pd connected to an OP-1 over USB. I'd love to know what this means and how to fix it.

It looks like I'm actually running into the issue described here: http://lists.puredata.info/pipermail/pd-dev/2013-06/019499.html

It would seem that the assessment there is correct: that pd is not correctly handling sync data, and a buffer is overflowing somewhere. Like the person who sent out that email, I haven't been able to get portmidi to build for OS X either, and so I haven't been able to attempt to rebuild pd with any kind of fix.

(Also, fwiw, the translation appears to be "Real-time Scheduling" )