incorrect handling of long (multi-packet) field?

[Wim Lewis]

I'm not sure if this code is actually incorrect, or if it just looks wrong, but in field.cc near line 287 there's this code:

    {
      con->pop_state();
      con->push_state(drizzle_state_packet_read);
      con->push_state(drizzle_state_field_read);
    }

The intent, I think, is to cause libdrizzle to read the next packet header and then return to reading the current field, but the two 'push' operations are in the wrong order for that.

(Since the current state is presumably drizzle_state_field_read anyway, I think both the pop_state() and the push_state(drizzle_state_field_read) could be removed, leaving just the push_state(drizzle_state_packet_read) --- after which libdrizzle would return to what it's doing now, i.e., continuing to read the multi-packet field.)

[Andrew Hutchings (linuxjedi)]

you are totally right, I think that is really old code. Good spot. Removing those worked perfectly and I believe gave us a nice small optimization too :)