Hello to all, I have problems using precompiled boundaries. It does not work to me. Do I understand it right: if you have a directory "bounds" with boundaries then mkgmap options "--location-autofill=bounds --bounds=bounds" should set the style file variables mkgmap:admin_level2, mkgmap:admin_level4 and so on? I've tried both generated by me and taken from navmaps.eu. I don't have any error messages or alike but It seems mkgmap cannot find the boundaries and the variables mkgmap:admin_levelx are always empty. What works for me is getting data from addr:xxx. But if I comment out the corresponding rules (so that there are only mkgmap:admin_levelx rules active) my elements don't get city and country set, for example. The "bounds" directory is in my working directory where I compile maps. The problem is not corrupt boundaries or similar. They are ok. Mainly I work with my home town Bad Düben, Saxony, Germany. I've read a lot here on the list but it doesn't really get me forward. I use last version of mkgmap (r2328), but the former version did not work for me either. As style I use the default shipped with mkgmap. In one thread I saw a tip of WanMil to look at the log and search there for strings like "N 617233095 ;;;;;;Bezirk Innsbruck-Stadt;;Tirol;;AUT;;" but I don't have in my log file such a text or anything else related to LocationHook. Probably I miss something very stupid but can't see it. Any advices are very appreciated. Maxim Düster
Maybe you didn't use --index in your mkgmap arguments? I think this was also the missing link in Lambertus' maps with a not working-as-expected index. Maybe --index should made default when location-autofill=bounds is used?
Maybe you didn't use --index in your mkgmap arguments?
Yes, exactly, that's the point! As I thought, it was something simple. Actually I've had the index option as argument but only for the second call, when I put all tiles into one gmapsupp.img. So that was my fault. Thank you Minko! :)
Your welcome, I was dealing with exactly the same problem and it took me some time to figure out there was that option --index missing too, see http://forum.openstreetmap.org/viewtopic.php?id=2625&p=66 So maybe it would be better when this --index would be by default turned on in the --location-autofill=bounds call?
Your welcome, I was dealing with exactly the same problem and it took me some time to figure out there was that option --index missing too, see http://forum.openstreetmap.org/viewtopic.php?id=2625&p=66
So maybe it would be better when this --index would be by default turned on in the --location-autofill=bounds call?
Up to now we do not have rules in mkgmap that an option is set if another option has a specific value (I think so?!?). Do we want to have such a coupling of parameter values? I prefer to have an error message if location-autofill and/or bounds is set but index is not. I have started to change the implementation of the autofill parameters (read http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2012q3/014885.html for details). It would be good if you contribute so that the new parameter handling is more understandable than now. Thanks! WanMil
Hi Why is the index option required? No reason that I can think of. The test that prevents the bounds being used when the index option is not set should just be removed in my opinion. ..Steve
Hi
Why is the index option required? No reason that I can think of. The test that prevents the bounds being used when the index option is not set should just be removed in my opinion.
..Steve
You are right. It is not required. Anyhow it is very likely that a user wants to create the index if the bounds option is set. When setting bounds but not index one is able to use location specific style rules (e.g. use country specific types for highway=motorway or highway=residential). At least a warning should be reasonable. But we don't have a good place for warnings. The log system is rarely used and it must be configured to show warnings. Stderr should printout errors only?! WanMil
Hi
Why is the index option required? No reason that I can think of. The test that prevents the bounds being used when the index option is not set should just be removed in my opinion.
..Steve
You are right. It is not required. Anyhow it is very likely that a user wants to create the index if the bounds option is set.
When setting bounds but not index one is able to use location specific style rules (e.g. use country specific types for highway=motorway or highway=residential).
At least a warning should be reasonable. But we don't have a good place for warnings. The log system is rarely used and it must be configured to show warnings. Stderr should printout errors only?!
WanMil
Ok, I have jumbled some things. Sorry. The index option only defines that a *searchable* address index is created. Without the index it is still possible to assign an address to each POI. And at this point it makes sense to set the bounds option. So it makes no sense to issue a warning or an error if bounds is set but index not. Another question: After reading the source code I think that a POI must have a city name. It is not possible to set the country and/or region without a city name. Is that correct? Should we use an empty "" city name for such cases to be able to assign the region and country info to a POI? WanMil
On 07/09/12 19:50, WanMil wrote:
After reading the source code I think that a POI must have a city name. It is not possible to set the country and/or region without a city name. Is that correct? Should we use an empty "" city name for such cases to be able to assign the region and country info to a POI?
Thats true: A POI (or street etc) is in a city A city is in a region OR in a country A region is in a country. I don't think a totally empty name would work, but you could certainly try using a name such as 'unknown' or something. ..Steve
On 08/09/2012 11:41, Steve Ratcliffe wrote:
A POI (or street etc) is in a city Not necessarily - plenty of them are monuments, natural features, rural post offices, etc, who would be most offended to be considered to be in a city. But I agree a placeholder might be necessary.
Roger
_____________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
-- ------------------------------------------------------------------------ Roger Calvert ------------------------------------------------------------------------
On 08/09/12 12:32, Roger Calvert wrote:
A POI (or street etc) is in a city Not necessarily - plenty of them are monuments, natural features, rural post offices, etc, who would be most offended to be considered to be in a city. But I agree a placeholder might be necessary.
I'm sure they would, but I was talking about how the Garmin format works. ..Steve
Hi
You are right. It is not required. Anyhow it is very likely that a user wants to create the index if the bounds option is set.
Yes, although not necessarily in the same run. I usually create the tiles first and then the index. But then I am probably not a typical user. Even if you are not creating an index, the --bounds option is still useful and results in a better map. The warning should almost be the other way around, creating an index without using a bounds file leads to a poor index.
At least a warning should be reasonable. But we don't have a good place for warnings. The log system is rarely used and it must be configured to show warnings. Stderr should printout errors only?!
Agree, we shouldn't use the logging for warnings. Although in fact currently there is a mixture of log.error() and System.{err,out} in use for errors and warnings. I'd prefer to use System.err for user messages with the message prefixed with "Warning" if appropriate - one day someone may be motivated to create system to print messages with internationalisation. ..Steve
Hi, I think it should be possible to get mkgmap:country etc. also if no index is needed. Henning
participants (6)
-
aighes -
Maxim Düster -
Minko -
Roger Calvert -
Steve Ratcliffe -
WanMil