reset_timestep bug

There is a small bug in Output::reset_timestep which can cause the
value of next to become undefined if no dumps are set up.
next_dump_any will be undefined in that case (and so will be all
values derived from it).
The patch below sets next_dump_any to MAXBIGINT (all operations below
involve taking a minimum, so if dumps are defined it bahaves well, and
if not it won't affect the value of next either)

diff --git a/src/output.cpp b/src/output.cpp
index 8237ab8..d07e18a 100644
--- a/src/output.cpp
+++ b/src/output.cpp
@@ -449,6 +449,7 @@ void Output::write_restart(bigint ntimestep)

void Output::reset_timestep(bigint ntimestep)
+ next_dump_any = MAXBIGINT;
   for (int idump = 0; idump < ndump; idump++) {
     if (every_dump[idump]) {
       next_dump[idump] = (ntimestep/every_dump[idump])*every_dump[idump];
@@ -464,8 +465,7 @@ void Output::reset_timestep(bigint ntimestep)
       next_dump[idump] = nextdump;
- if (idump) next_dump_any = MIN(next_dump_any,next_dump[idump]);
- else next_dump_any = next_dump[0];
+ next_dump_any = MIN(next_dump_any,next_dump[idump]);

   if (restart_flag_single) {

this fix was included in today’s patches …