On Aug 13, 2009, at 12:34 AM, Michael Richter wrote: > I've isolated the problem. (Not solved it, but identified it.) > > The problem was network congestion somewhere between China and > wherever sqlite.org is located. After I did a complete shutdown of > everything and restarted from scratch (renegotiating my PPPoE lease) > the problem went away by itself. Further what was taking literally > hours before only took a few minutes this time. > > So... > > It seems that Fossil is being a little bit optimistic about the > network's reliability. Given the random locations that things were > dying in, I'm guessing there was network timeouts that weren't being > gracefully handled (as in reported) causing Fossil to process > incomplete sets of cards.
The code in http_socket.c treats a network error as an end-of-file. D. Richard Hipp [email protected] _______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
