[OSM-dev] Student Project Ideas?

[Steve Hosgood]

Another idea:

This would probably be best thought-of as a feature for API 0.8, but 
basically the idea is to "fix" the current sloppy formatting of the 
database tags that contain measurements. This would include "maxspeed=", 
"maxheight=" and "maxwidth=" but there are weight limits too, and 
"est_width=" and other such tags.

The "problem" with 0.7 and earlier is that the data in these fields is 
free-form and this doesn't suit the needs of autorouters. The U.S users 
and a lot of U.K. users are inputting these values in imperial units, 
everyone else is doing it in metric units. Autorouters need to work in 
some consistent system and are currently having to parse out the 
freeform data.

That's not too difficult for metric, but (say) a height limit in 
imperial may be stored as <<11' 6">> or <<11ft 6>> or <<11ft, 6in>> or 
<<11.5ft>> or.... you name it! And human users of editing programs might 
like to work in imperial-only (and see values that were entered in 
metric be converted for them) or metric-only (and see values that were 
entered in imperial be converted for them) or have it displayed the way 
it was entered.


I suggest that API 0.8 would specify that any values in the database be 
stored in some appropriate canonical form, with a flag to say if it is 
naturally imperial or naturally metric. So heights and widths would be 
stored (say) in integer millimetres or integer inches with a one-bit 
flag to say which it is, and speed limits would be stored in integer 
km/h or integer mph with a one-bit flag to say which it is. The API call 
to obtain these values would have a parameter to say whether the user 
wants metric, imperial or "how-it-was-specified".

( The reason for the one-bit flags is to avoid converting data when it 
needn't be. Autorouters will certainly want to work in a single given 
system and won't mind if a speed limit is 48.27kph, but humans would 
probably rather see that one displayed as 30mph - even if they are 
French! Likewise human U.K users would probably want to see a French 
speed limit shown as 50km/h not 31.0752mph. )

The GSoC project would need to deal with writing a robot that could go 
through all the existing data, parse the extensive list of inventive 
abuses of the free-format data(!) and rewrite it in this canonical form.

The project would also create this new API call to return the data of 
course. Presumably it would present data in human-readable form 
according to the "how I want to see it" parameter.

Steve.