... is that there are so many to choose from... well, in the case of DCC (Digital Command Control) for Model Railways there is really only one, the "official" standard as described by the NMRA in the US.  DCC has been around for many years - probably something like 15 or more, so you'd think it was pretty established by now.  The fact is that things haven't progressed that far, with equipment very expensive and quite unintuitive to use it's really not gone beyond those who were both computer-literate and model railway fans, even though it doesn't require a computer necessarily the level of complexity to understand configuration variables, registers and so forth puts most people off without a second glance.

So here we are, coming in to 2007 and here in the UK we've got both Hornby and Bachmann; who represent the train set market almost entirely between them, releasing DCC systems.  Prices are coming down all over the place, you can get decoders for as little as about £8 (approx $16) - which is great compared to the £30 decoder cost from when DCC really started over here.

It sounds like everything is great, we finally have a digital system that's standards compliant and everything is good...

Sadly, it isn't.  It's far from that, and getting further with each new system that's launched.

Why? Because the manufacturers are simply not putting their equipment forward to the NMRA for conformance testing! They run informal tests with one or two other vendors equipment (Hornby say they tested their set against the Lenz equipment for example), but to use an analogy, just because I can find a way to talk to a French person does not mean that I can talk French.

What's happening is that little incompatibilities sneak in and you get odd occurences of particular combinations that don't work or exhibit odd behaviour.  The Gaugemaster Prodigy (which is just a rebadged MRC device) is one example - it can't talk to the Lenz Silver Mini decoder at all, and has problems talking to the Hornby decoder.  In the case of the Hornby decoder they are doing an update to the controller to add support for it, but why does a controller have to be upgraded to support a particular vendors decoder? That's what the DCC standard is all about!  Call me a sensationalist, but in my view this is rapidly turning in to a situation where we have lots of vendor specific equipment that does not guarantee to work with anyone elses and might, or might blow it up (I hear rumours the Hornby Select sometimes peaks out at extreme voltages outside the official standards acceptable ranges that might well destroy some decoders!), or might not work, or might not work quite properly, if you want to be sure it works then you have to stick with one vendor... but to me, that wipes out one of the major reasons for going to a DCC system and will put in serious jeapardy how well it gets adopted in the UK.

A little background...

Hornby already did Digital many years ago with a system called Zero-1.  It worked great but had lots of limitations, the biggest was that it was Hornby proprietary.  Airfix had a similar system as did one or two others - but none of them made any efforts to communicate with any of the others.  The systems all died a death because people didn't want to get locked in, or they bought the equipment and it got outdated, the only way forward was to simply go back to the old way of doing things.

While we were at the Leamington and Warwick show, the single biggest question asked of us was : "Is this another Zero-1?" or questions alluding to the same thing.  Of course with the NMRA conformance warrants and associated processes the idea is that no, it very much isn't, it's all compatible between vendors and "its all good"...

... but take a look at the NMRA Conformance Warrants list and see just how few pieces of equipment are on it, particularly from the UK.  That's right, Hornby and ZTC don't figure at all.  Lenz? Well, they make a good showing but their Compact controller is nowhere to be found.  Bachmann US are great, very pro-active and presumably have an advocate of the process in their organisation; but Bachmann UK are much like any other UK organisation.  Even more strange, while I was checking the documentation for a Bachmann 36-553 decoder (which does not figure in the conformance warrant list by that number at least) is displaying both the NMRA Conformance 'football' logo and the NMRA DCC logo - which is meant to imply it is NMRA DCC Conformant... so what's the truth? Are the logos being used incorrectly? Is this perhaps a rebadged alternative decoder?  The issue i'd have there is that it's a 3 function decoder that uses short addressing - the manufacturer ID is that of ESU, but the only ESU decoders listed as warranted are the LokPilot's - which are four function long addressing decoders.

It's a right old mess, and it needs pro-active action to sort it out before it destroys DCC in the UK.

The Saturn platform is progressing well.  If you don't already know, Saturn is the name of the platform that's being developed to power the new versions of UKTrainSim, UKFlightSim, UKTruckSim and UKShipSim.  There's a whole new professionally designed look and feel to the sites, and the Saturn platform architecture is revolutionary compared to the current rubbish - hence the name, it runs rings around what we have now... I know, my sense of humour needs work.

As at this moment much of the download library support is there now, aside from the ability to actually download or upload something, but I don't let myself get hung up on trivialities :)  Getting the dynamic data system working was good, you can now have unique attributes on each different kind of file, so a Flight Sim mission can describe its difficulty, the aircraft you'll be flying and how long it is, a Train Sim route can describe it's period, the length, whether it's express or local and so forth - rather than trying to squeeze it all in a "one classification fits all" system the way that UKTrainSim currently works.

I've got the navigation of the new folder system working too, again, where UKTrainSim currently has a rigidly fixed two-folder system (file type and operating company) the new one can go as many levels deep as you like and the same files can be in multiple folders so that you can find them by various paths.

The parts of the site for trivialities such as Editorials, Articles and Reviews are done, the news system (at least reading) is done and the front page is essentially finished - and that's the most complex page of the lot since it draws on just about all the sub systems in the new architecture.

One of the key goals of the new system is to improve scalability and reliability, the current system isn't scaling so great - it's running on pretty big iron to perform at the moment where the new system should scream on the same platform.  Similarly it's just not very reliable, all things considered, with Apache or PHP (99% of the time PHP) exploding and causing the server to have to restart it's just not very good.  I'll be glad when PHP can be chucked in the bin, where it belongs, personally. 

For those interested in technobabble, the new platform is based on Java 5 Standard Edition, but also calls on some J2EE facilities such as JSP.  It utilises Struts 1.3 as its application framework (Struts 2 is not released yet) with Tiles to enable componentisation of the pages themselves (which is outstanding!).  It makes use of Hibernate and Hibernate Annotations in order to access the database, which improved development time considerably, I spent hours and hours writing all the business object model for the current site in the PHP code, and it was done and tested in a couple of hours (including learning Hibernate itself) with the new system.  I'm also making use of a number of caching algorithms to optimise access to the database, this should make it a lot more performant as long as I can get the strategies balanced as far as reading/writing are concerned - ideally at some point the system should scale on to multiple front end servers, it's currently a single server for front end and a single server for all database.

There is also a totally new database design for the new platform, there were a number of obviously brain dead design factors in the current version that UKTrainSim operates, though they haven't come to light until the system hits considerable load it does highlight the need to design a system with high load right from the outset... then again, to be fair to myself, I had never anticipated the system being quite this busy!

There's still a ton of work to be done to get the system ready to launch, but the jobs are being whittled away gradually and I feel that it's definitely on the downward slope now (finally).

A lot of the functionality that I had intended / hoped to put in to the new version from the outset won't be there, but it will certainly be a number of steps beyond what we have now and on a platform I am happy to continue adding to over time.  Part of the challenge will be seeing how people get on with the new approach to things, hopefully they will appreciate the extra flexibility; but flexibility is always on the other side of the see-saw with usability, so that's going to be a challenge to balance - hence, get the system online as soon as it's capable and then look to get user feedback and enhance or alter the system accordingly.

Later I'm aiming to build a "web services" interface to the site so that other 3rd party tools such as Route Riter, Train Store etc have the option of integrating directly to the site and in essence becoming "clients" for people to use, automatically downloading missing files and so forth.  Depending on how things go with new simulators (Kuju and Microsoft's new ones) there's also a desire to provide some automatic installation and content management tools as well, but we'll have to see - that's a fair way down the tracks and we need to take this one tie at a time :)

Matt.

Matt.