On Tue, Sep 2, 2008 at 11:08 AM, spaetz <[EMAIL PROTECTED]> wrote: > On Tue, Sep 02, 2008 at 10:08:11AM +0100, Dave Stubbs wrote: > >> I think we had this discussion before and came to the conclusion that: >> - 50mph was essentially mapping a sign, because the speed limit is a >> speed, not a unit ... > > And as some applications might want to show the precise sign value, not some > rounded appoximation (agreed that apps can round).
I've never seen a speed sign for anything but integer units. Given the accuracy of most speed measuring devices I'm guessing I never will either. People rarely set speed limits at anything other than divisions of 5 units be that mph or kph, unless the result is actually a conversion. ie: the speed limit in Windsor Great Park is signed as 38mph. So if I convert 112kph to mph, and round... the result is not an approximation, but the actual signed speed limit. This is one of those know your domain things. > Plus it's more intuitive for the mapper. Agreed. > >> - that the tag without a unit should probably be assumed to be km/h. > > that's why my example used maxspeed=50mph and maxspeedd:mph=50 in case you > haven't noticed :-) > >> - that anything intelligent enough to know if it wants to represent >> maxspeeds in mph/kph is intelligent enough to know it can safely round >> to the nearest integer. > > Anything being able to round to the next number should als be able to read > miles (or have a clever enough preprocessor to do it :-)) > >> - and that it's possible to represent an exact mph in kph anyway if >> you can really be bothered: 1mile == 1.609344km exactly > > Do you always carry your calculator with you when mapping or do you do it by > hand :-) > > I am not saying that it shouldn't be tagged as a rounded km/h value. However, > people shouldn't think they are forced to. If they feel that maxspeed:mph=50 > makes more sense, than that should work too. > I'm actually just summarising the mess that the last time this discussion came up, where I was actually arguing the mph case. http://lists.openstreetmap.org/pipermail/talk-gb/2007-September/002417.html About half of the arguments are a little weird, not least the whole rounding thing, which it turns out really isn't a problem in any sensible application. Dave _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk