Bob and Dave,

Since you both asked after my plans for the future of my railway, here's the summary of where I want to go with my layout. Or rather, layouts.

What I want to do has intermediate stages. Although I have in the past already successfully connected to a tiny circle of track and a program track (NCE PowerCab and NCE USB) and programmed all decoders to my satisfaction, the first stage is entirely virtual. There will be an indoor layout consisting of a single track circle with three passing places. No further frills at all. That's all measured out, and exists in JMRI as a Layout Editor panel with a Panel Editor panel for each of the three passing places. So six switches/turnouts/points. Signal system in place, everything working fine under simulator and everything (so far) virtual.

Eventual goal is a more complicated garden railway (still nothing fancy, but more stations and more involved than the extremely simple test track). Type of operation is almost exclusively passenger trains, occasional goods train in between, but fixed rakes of wagons, so no switching other than engines being replaced by others or running around at a terminus (it will be point-to-point). To that end, all locos have DCC-controlled automatic couplings installed.

Stage 1: that's where I am - JMRI able to control my indoor layout that doesn't exist yet. All the hardware for stage 2 is available (rails, locos, PowerCab, NCE USB, detectors, two AIUs, switch decoders) and a lot for stage 3 as well (though no Power Pro yet).

Stage 2: Actually build the virtual layout and make JMRI control it all. Two trains in opposing directions, halting at all stations, totally automatic. Right-hand traffic. Trains to run autonomously, but heeding (virtual) signals (more on this a bit later). Single track sections secured by (virtual) signal boxes. Each end of each station has one signalbox, that communicates with the one in the next station. In tandem, they prevent cornfield meets, each controlling their own switch and the signal (only held/not held) of a train departing into the single track to the next station. Signalling (signal masts, my further simplified version of 'basic' signalling) is working, the signalboxes and the state machines inside them was where I was going to use python. That makes six instances of identical signalboxes (I love OO). No timetabling; each train just runs as and when it can. Apart from the signal boxes, I'll also need a 'driver' object, able to control a JMRI throttle with knowledge of the way around, speeds, time it takes, distance to next signal, and ability to 'see' the appropriate signals.

Stage 3: Expand on this into the garden, with more, and more complex, train movement, at least one larger station, timetable control (an additional OO object that both signalboxes and drivers need to be aware of) and a 'plan' for the driver to follow. Again, I mean to have the signals control train movement rather than the other way round, even when the signal system already works by being controlled by the trains. So: playing with 'held' alone. Add into that mix the ability to take control of any train myself.

Quite ambitious, I think, but nothing undoable as far as I can see.

There's also a live steam loco ready to go, but mixing that in isn't something I'm really prepared to think about yet. It has no decoder (and don't plan one) but at least the wheels are isolated.

Enough to keep me ging for a number of years to come!

Thanks again, both, for your time, patience and insights.

Wouter

On Sat, 5 Jan 2019 at 21:10, Bob Jacobsen <rgj1927@...> wrote:

There are quite a few developers who work via JMRI via Eclipse.? The Eclipse support (see ) occasionally has rough edges because Eclipse evolves pretty fast.? If you’re experienced with it, though, I think it’s probably a good entry point.

Most of the instructions are about working with the JMRI code itself, which isn’t entirely required.? You could just include the jmri.jar and work with it as s formal library, but pulling the while thing gets you (a) an environment that a bunch of people know works and (b) easy access to Javadoc, etc, within Eclipse.

(Even with Java code, a bit of scripting can be the easy way to invoke your new classes, etc, so that might still be a bit on the table.)

Could you remind me of what you’re working on as your first project?? I can perhaps point you to some examples and docs that’ll help you pull together a starting point.

Bob

> On Jan 5, 2019, at 12:55 PM, whmvd <vandoornw@...> wrote:
>
> I will ask for more advice, though. And that is: what's good tooling for me to start out with (on Linux)? I'm familiar with Eclipse, but not so in-depth that I'd hate to use anything else. And what are gotchas with how I set up my coding bit separately enough so that a new JMRI-release slots in nicely and easily?

--
Bob Jacobsen
rgj1927@...