[OSM-dev] Drawing ways with bezier curves rather than linesegments

Dave osm at randomjunk.co.uk
Thu Jan 25 15:41:28 GMT 2007

Jeroen Dekkers wrote:
> At Thu, 25 Jan 2007 10:03:01 +0000,
> Dave wrote:
>> Andy Robinson wrote:
>>> I'll stand corrected on this but I recall we had a long discussion on bezier
>>> curves maybe more than a year ago, a search of the list (probably on talk at
>>> the time) should find something. One of the issues that was flagged at the
>>> time was a patent that Navtech has on bezier curves. I think it was one of
>>> the reasons the idea was dropped quicker than it might otherwise have been.
>> You can't patent bezier curves, there's prior art. Saying we will use 
>> bezier curves, /but on a map/, is such a non-patent it isn't even funny. 
>> And automatically figuring out what curve to use from a series of 
>> straight lines has been functionality in vector graphics programs for 
>> donkeys years.
> Prior art doesn't really stop the patent office from granting
> patents. 
Very true, although that's just laziness really as it does invalidate 
the resulting patent -- if you can afford the lawyers to challenge it 
that is.
>> Keep the software in Europe and for the moment at least we avoid the 
>> insanity of requiring an army of lawyers and the backing of a 
>> multinational to do things which are basic and obvious.
> Most software patents exist in Europe too. For example, EP0919788
> (https://publications.european-patent-office.org/PublicationServer/xmldocument?docId=4767068#CLAIMS)
> is the European version of US6366927.

I think my point is I'd rather nobody on the list knew anything about 
any patents that might or might not be related to something we're doing. 
Mostly this is because:
a) Half the time we don't know what the patent is actually for
b) We have no real way of assessing whether or not we're infringing
c) We have no sensible way of determining whether, if we are infringing, 
the patent stands a chance of being valid

If the first thing we know about something is someone actually making a 
fuss about an infringement, then we stand a better chance of arguing 
that the patent was stupid to start with. We can then either chose to 
fight or change what we're doing. But as the truth is we're almost 
certainly going to step on a million patents just by storing lat/lon 
points on a computer, and doing it using some computer programs, and 
we're never going to try an exhaustive search to check we're not 
infringing (no-one does as you'd need infinite resources), there's very 
little point in occasionally asking the question about something someone 
stumbles across.

One software company I've worked for had the policy that the only 
interaction developers should have with patents is that if they think 
they come up with something new they should tell the lawyers. The 
lawyers then ask some questions. The developer may get told at a later 
date that they now have a patent, yey!, which came with a bonus in the 
pay packet. It's very explicit, that at absolutely no point should a 
developer try and figure out if something is already patented, their 
absolute best state is ignorance, and they should /never/ ask the lawyer 
what happened to that patent idea they had last year...
But then this company has a massive patent portfolio to smash people 
with if they do something they don't like. They also have more lawyers 
than most companies have total workforce.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20070125/a71f1350/attachment.html>

More information about the dev mailing list