version 0.9.8 released

Dear users,

Again, it's time for a new SUMO-release (0.9.8). The list of changes is available at http://sumo.sourceforge.net/wiki/index.php/ChangeLog#Version_0.9.8. Before listing the most important ones, we would like to thank all those who have contributed by sending bug-reports.

This will probably be the last 0.9.x-version of SUMO. In the next time, we will try to adapt the changes in the API as proposed in http://sumo.sourceforge.net/wiki/index.php/Specification. This will affect the behaviour of the simulation in several parts and change some input and output files. Because several people are co-developing SUMO, we would propose to discuss these changes on the developers-mailing list (sumo-devel@lists.sourceforge.net).

If you are not interested in development, we still encourage you contribute to the users-mailinglist (sumo-user@lists.sourceforge.net) and to report your problems or thing you found odd while using SUMO.

The following changes from the prior version seem to be the most important ones to us:

- The structure of the project was reworked; many subprojects (modules) were joined and the directory structure is now resembled within the msvc8-project

- debugged vehicle emission bug: vehicles with a depart time lower than the simulation begin time were emitted, too. Definitely a large bug

- debugged behaviour on junctions; vehicles no do not halt some meteres in front of them but on the correct halting position; this affected simulation without inner-lanes only

- some further work on inner-junction traffic; looks very fine, but one thing is still open: the vehicles do not regard whether they have place on the consecutive lane.

- vehicles now do not decelerate on higher priorised roads; finally, this was quite a large bug

- the simulation loop was reworked. By computing the real leader/follower - meaning that also predeccessor/following lanes are taken into account when needed - instead of the old procedure of "signalling" that a vehicle approaches a lane we were able to a) make the simulation faster because we do not have to touch all lanes in each step and b) make the information about leader/follower more explicite

- complete detectors refactoring; making the implementation easier to understand; tests added

- a preliminary OSM-network import

Bye,
Michael & Daniel

Posted by Michael Behrisch 2008-02-07