[Talk-us] I've been workin' on the railroad (in California)

stevea steveaOSM at softworkers.com
Fri Jan 16 04:38:41 UTC 2015

And here is what I've learned.

This is tedious work, especially a state at at time.  It is slow, but 
that is because it is a whole (large) state.  This is not impossibly 
large work, not by a long shot.  Yet for one person, a whole state is 
a lot of work.  For one railroad (elephants are eaten best one 
forkful at a time) this sort of effort is not difficult, and the 
rewards in new renderings are available "tomorrow."

ORM (http://openrailwaymap.org) is useful to see changes, though it 
does have a 24 hour (or longer) render time.  Primarily, the usage= 
tag (usage=main or usage=branch, which render orange and yellow, 
respectively) is useful, though other tags and colors are used.

It is important to use the name= tag to unambiguously (assert) a name 
for an aggregation of rail segments into a Subdivision or Line (or 
Lead if usage=industrial or service=spur).  This is especially true 
as OSM now often has messy rail around yards and regarding whether 
there is single-track or double-track in many or most places.  What I 
am saying is that we work to do to be more accurate.  And, this work 
is doable.

Both tags on ways and tags in relations are supported, but it appears 
tags on ways supercede.  I find that in the USA, because of the way 
that TIGER tags have put much rail into parts of North America, 
adding usage= tags to ways is often how things begin.  Then these get 
aggregated into type=route, route=railway relations.  I have not yet 
explored the multiple possible options of how tagging might supercede 
except as above.

That said, a much richer branching structure certainly exists 
(requiring appropriate tagging).  Sidings, yards and signalling are 
crude or virtually nonexistent, though there in some measure.  Speed 
limit data are virtually nonexistent.

Private corporation rail data are not likely compatible with our 
ODBL, though it doesn't hurt to ask (for explicit permission). 
Public rail data from a state with "generous" public record laws 
(such as crossing spreadsheets) can be quite helpful to identify 
names (based on road crossings, city/county and/or lat/long data) of 
Subdivisions, Lines and Leads.

This is slow going, but I do appreciate the explosion of activity I 
(and others) can see.  Slow and steady does improve, and I nod my 
head at how we have both good data and new activity.  Railfan 
crossover contributions are welcome (and I know are out there).

Perhaps a WikiProject would be helpful (if no more than a place to 
meet and chat).  And:  so far, so good.


More information about the Talk-us mailing list