[mkgmap-dev] [PATCH v5] - merge nodes to remove evil short arcs

Thu May 28 22:46:50 BST 2009

Hi Marco,

> I compiled Italy right now with the last jar version you sent me (1044 + V5, I guess).
>
> GREAT!
> 
> I've tryied to crash MapSource without any success (so far...): all roads I kew were problematic are now routable without issues.

That's great.

> For your info, in Italy about 30/50 arcs seems to be "unroutable" (I'll try to correct them as soon as I have the OSM ID info).

It's possible that the code that detects self intersecting ways is
underperforming and it should be fixing those ways but is failing. So
far, the only ways that I have seen it complaining about do have silly
connections so maybe it's not a big problem. If you find any ways that
really should be routable but are not, let me know.

> What to say more. Well done. As soon as you are confident to commit the patch in the trunk (I guess this is the right way to call what I called the mainstream mkgmap), I'll try the last version.

Well, you should take a good chunk of the credit as you put
time+thought into sorting out what was wrong.

I will commit this soon (possibly tonight). As you have to specify
an option to turn it on it will not cause any problems for people. In
the future, it may be enabled by default but let's give it a trial
period first.

> Well, just a question: why if I put a wrong switch (like --delete-short-arcs) mkgmap does not tell me there is an unrecognized switch? I might use a wrong switch without knowing...

Yeah, it's a bummer but mkgmap doesn't keep track of what options are
valid - the options are detected in whatever part of the code is
interested in them but if you specify a bad option name, nothing
notices. Should be fixed.

Cheers,

Mark