A new version of the release candidate for Bernd Gabriel's helpful work on merging the latest stable version of Standard into the latest version of Experimental, Simutrans 112.2 Experimental 11.9003, is available here (http://bridgewater-brunel.me.uk/downloads/Simutrans-Experimental-RC.zip). This is just the Windows binary, but it should suffice for testing purposes.
Edit: I have updated the .zip file to include a single threaded version lest the multi-threading was responsible for any of the bugs that have been reported and which I cannot reproduce.
It is based on version 112.2 of Standard (the latest release), merged into Experimental by Bernd, and work that I have done in the last few months to reform the routing of private cars (to make it workable on larger maps: see here (http://forum.simutrans.com/index.php?topic=9595.msg111178#msg111178) for a discussion on this) and also the work to improve the simulation of passengers walking (see here (http://forum.simutrans.com/index.php?topic=11254.0) for a discussion of that feature).
It fixes a number of bugs from the previous release candidate (thank you to all those who reported them, and to Nathaneal for fixing some of them) as well as incorporating a slight change to the way in which normalised random numbers are generated as suggested some time ago by Kieron which will hopefully give a better range of journey time tolerance figures, improving passenger generation somewhat. It also fixes some undefined behaviour in 11.9002, which has the potential to make online games more stable.
Version 11.9001 introduced a new way of handling weight limits: from that version onwards, weight limits on railways and roads of all sorts are be based on the axle load, which can be set in the .dat file with the axle_load=X parameter, or, with the latest version of Makeobj (people will have to compile it themselves for the present), the axles=X parameter (only to be used where the axle load is evenly distributed). The axle weights do not fluctuate with the convoy's load. Bridges will instead use the total convoy weight, which will fluctuate with the convoy's load, but this total load is shown in the depot/replacer window before dispatching the convoy. Note that the length of the convoy relative to the bridge is taken into account, so that only the maximum number of vehicles in a convoy that can be fully on the bridge (excluding the slope tiles at each end) will count towards the weight. Suppose, for example, that a convoy is 200 tonnes in total, but 5 tiles long: if a bridge, excluding end slopes, is 2 tiles long, only 2/5ths of that convoy weight (80t) will bear upon the bridge at once, so a bridge with a length of 2 tiles and a weight limit of 80t will be sufficient to take the load. This should substantially enhance the realism of way based weight limits. I am currently working on revising the ways in Pak128.Britain-Ex ahead of its next (and long overdue) release, and this change was intended to facilitate that process.
I should be very grateful for all testing of this version, which will hopefully (with whatever modifications that such testing shows are necessary) become the next major release of Simutrans-Experimental. I should particularly appreciate feedback on the new features relating to passenger walking, the use of private cars and weight limits, as well as the usual feedback on stability and performance. Please post a separate thread for each bug or topic/issue of feedback.
To make the most of the passenger walking feature, I suggest that players set their station coverage radii to a much higher than normal value (in the simuconf.tab file of the pakset used): a value of 16 is recommended.
Thank you all for testing - it is much appreciated.