
Hi, I successfully created OSM maps of great britain for my Edge 705. I create the img file with --index and use bounds either I created myself using osmosis or from precompiled (from navmaps.eu). The Edge 705 loads the img map correctly and address search is working properly. However, when accessing the function "Cities", which lists the closest cities, the Edge freezes for 1 or 2 minutes before displaying cities (ordered by distance from current location). What I noticed is that between some cities names there are also blank lines, as if there were blank entries in the cities list. Scrolling the cities list (which may be possible after 1-2 minutes of freeze) is faster when there are none of these blank lines, which may suggest they are responsible for the problem. But when one of these blank cities appears, then the Edge freezes again. Disabling --index solves the problem, but address search is not available then. Furthermore, some cities are listed under GBR, but many others belong to "Country" with a tag "ABC". Is that an OSM tagging problem? Has anyone noticed this? I would be happy to help diagnose the problem, what is your advice? Many thanks, Eric

On Fri, Feb 17, Eric Fernandez wrote:
Furthermore, some cities are listed under GBR, but many others belong to "Country" with a tag "ABC". Is that an OSM tagging problem?
Do you use the --location-autofill option? I see the same without that option, and I think it because of missing boundaries of admin_level >= 6. I use "--location-autofill=bounds,is_in,nearest" to solve the problem. Thorsten -- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

2012/2/17 Thorsten Kukuk <kukuk@suse.de>:
On Fri, Feb 17, Eric Fernandez wrote:
Furthermore, some cities are listed under GBR, but many others belong to "Country" with a tag "ABC". Is that an OSM tagging problem?
Do you use the --location-autofill option?
I see the same without that option, and I think it because of missing boundaries of admin_level >= 6.
I use "--location-autofill=bounds,is_in,nearest" to solve the problem.
Thorsten
Many thanks, I will try this. I use indeed --location-autofill, I tried --location-autofill=bounds and --location-autofill=1. Eric

I use "--location-autofill=bounds,is_in,nearest" to solve the problem.
Thorsten
Hi, this helped a lot using freshly calculated boundaries with osmosis. On the other hand, the precalculated bounds gave me bad results. May I ask you if you generate your bounds yourself, and if so, which settings/options you are using? Also, I still have the cities search bug on the edge, and can confirm the freeze occurs when a "blank" city is displayed on the list. The Edge lists the closest cities from your current locations and order them by distance. The display can hold 5 lines of city names. If there is no blank name on this list, then scrolling is fine and fast. But if one of these blank lines are on the screen, then the Edge becomes unresponsive for 30 seconds to 1 minute. Since it calculates the distance to each city, I guess it is probably stuck in computing an impossible figure. I am puzzled because I do not know how to correct this or to diagnose this, but am pretty sure this does not happen without --index. Can there be a bug in handling the index population? Thanks, Eric

On Fri, Feb 17, Eric Fernandez wrote:
I use "--location-autofill=bounds,is_in,nearest" to solve the problem.
Thorsten
Hi,
this helped a lot using freshly calculated boundaries with osmosis. On the other hand, the precalculated bounds gave me bad results. May I ask you if you generate your bounds yourself, and if so, which settings/options you are using?
I generate my bounds myself once a week from the complete planet file. You only need two options: --createboundsfile and --bounds. And of course a machine with SUN Java 1.7 and 8GB of RAM. Thorsten -- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

Hi, Following my recent report of my Edge 705 being temporarily frozen when scrolling nearest cities, I think I have narrowed down where this comes from. The problem occurs when I combine both --add-pois-to-areas and --index options. Firstly, the blank city lines are caused by --add-pois-to-areas. They do appear only when this option is set, so there may be a bug there. However, the freeze of the GPS device only occurs when --index is also used. Without --index, the blank lines are there but there is no slow down when scrolling in the cities list. Conversely, if I use --index without --add-pois-to-areas, then there are no blank lines, and there is no slow down at all. This happens with the Oxfordshire Geofabrik map, tested in South Oxford. Please let me know if and how I can be of more help. Kind regards, Eric

just a wild guess: if you use --add-pois to-areas, a lot of will cities turn up twice (one genuine poi and one geberated from the poly). maybe this srews up the index. to test it, you could use place=city & mkgmap:area2poi!=true in your point file and exclude cities generated by -add-pois-to-areas. Am 20.02.2012 14:47, schrieb Eric Fernandez:
Hi,
Following my recent report of my Edge 705 being temporarily frozen when scrolling nearest cities, I think I have narrowed down where this comes from. The problem occurs when I combine both --add-pois-to-areas and --index options.
Firstly, the blank city lines are caused by --add-pois-to-areas. They do appear only when this option is set, so there may be a bug there. However, the freeze of the GPS device only occurs when --index is also used. Without --index, the blank lines are there but there is no slow down when scrolling in the cities list.
Conversely, if I use --index without --add-pois-to-areas, then there are no blank lines, and there is no slow down at all.
This happens with the Oxfordshire Geofabrik map, tested in South Oxford.
Please let me know if and how I can be of more help.
Kind regards, Eric _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

2012/2/20 michael lohr <micha.lohr@web.de>:
just a wild guess: if you use --add-pois to-areas, a lot of will cities turn up twice (one genuine poi and one geberated from the poly). maybe this srews up the index. to test it, you could use
place=city & mkgmap:area2poi!=true
in your point file and exclude cities generated by -add-pois-to-areas.
Hi Michael, Thanks for your suggestion. I tried it but with no success, the problem still arises. Regards, Eric

Hi
Firstly, the blank city lines are caused by --add-pois-to-areas. They do appear only when this option is set, so there may be a bug there. However, the freeze of the GPS device only occurs when --index is also used. Without --index, the blank lines are there but there is no slow down when scrolling in the cities list.
This happens with the Oxfordshire Geofabrik map, tested in South Oxford.
I tried making a map with --add-pois-to-area from that map and didn't see any blank lines in either mapsource or on my Legend. So could be specific to the Edge or some other option you are using. Could you upload your map to http://files.mkgmap.org.uk (or anywhere), so I can examine it and see if there are any differences to mine?
Please let me know if and how I can be of more help.
Best wishes ..Steve

2012/2/20 Steve Ratcliffe <steve@parabola.me.uk>:
Hi I tried making a map with --add-pois-to-area from that map and didn't see any blank lines in either mapsource or on my Legend. So could be specific to the Edge or some other option you are using. Could you upload your map to http://files.mkgmap.org.uk (or anywhere), so I can examine it and see if there are any differences to mine?
Please let me know if and how I can be of more help.
Best wishes
..Steve
Hi Steve, I tried to upload to http://files.mkgmap.org.uk but after I press the upload button, the file path clears and there is no acknowledgement. The gmasupp file is 6.5MB big, is there a limit? Or is there a delay in showing the file on the page? Thanks, Eric

I tried to upload to http://files.mkgmap.org.uk but after I press the upload button, the file path clears and there is no acknowledgement. The gmasupp file is 6.5MB big, is there a limit? Or is there a delay in showing the file on the page?
You have to enter something in the description box too. There is an error message, but it isn't very clear. If it works, then it does go to a confirmation page after that. 6.5M will take a while of course. ..Steve

2012/2/20 Steve Ratcliffe <steve@parabola.me.uk>:
You have to enter something in the description box too. There is an error message, but it isn't very clear.
If it works, then it does go to a confirmation page after that. 6.5M will take a while of course.
..Steve
Hi Steve, This is done. My description entry was too short indeed. Thanks, Eric

This is done. My description entry was too short indeed.
OK, thanks. At first sight I can't see the problem, so it may only show up on certain devices. So about the blank lines: can you select them and show them on the map? If so what are they? Are there any clues as to what they are? How many are there compared to the number of cities? ..Steve

2012/2/20 Steve Ratcliffe <steve@parabola.me.uk>:
This is done. My description entry was too short indeed.
OK, thanks. At first sight I can't see the problem, so it may only show up on certain devices.
So about the blank lines: can you select them and show them on the map? If so what are they? Are there any clues as to what they are? How many are there compared to the number of cities?
..Steve
Hi, I can indeed select and show them on the map. When selected from the city list, they appear as little green diamonds. They are selectable on the map (they can be highlighted but there is no description associated). They are routable (I can ask the edge to go there) and then after a route going there is calculated, they appear as a "Point" that looks like a small round dot, identical to the small round dot is also used for "normal" city names/villages on the map. There is one at this location: N 51 42.620' W001 13.721' Another one there: N 51 42.522' W001 16.668' These blank cities are not very frequent, around 1 in 10-20 cities, but when they are indexed, they make the Edge 705 freeze. Thanks, Eric

Hi
I can indeed select and show them on the map. When selected from the city list, they appear as little green diamonds. They are selectable on the map (they can be highlighted but there is no description associated). They are routable (I can ask the edge to go there) and then after a route going there is calculated, they appear as a "Point" that looks like a small round dot, identical to the small round dot is also used for "normal" city names/villages on the map.
I might have found it, or at least part of the problem. In your map there are 10 cities without a name, created from polygons. Since there are around 500 cities in the map, this is closer to 1 in 50 than 1 in 10-20 so might not be everything -- do you see more than 10 or is it possible that is the total number of blank lines? It does include the two example you gave however. Both these are place=* polygons, with place_name instead of name. You probably can and should try to use the place_name tag to give the name, but I think that we should also prevent any unnamed cities from getting into the map file. ..Steve
There is one at this location: N 51 42.620' W001 13.721'
Another one there: N 51 42.522' W001 16.668'
These blank cities are not very frequent, around 1 in 10-20 cities, but when they are indexed, they make the Edge 705 freeze.
Thanks, Eric _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

On Tue, Feb 21, Steve Ratcliffe wrote:
It does include the two example you gave however. Both these are place=* polygons, with place_name instead of name.
That's why I use --name-tag-list=name,place_name meanwhile.
You probably can and should try to use the place_name tag to give the name, but I think that we should also prevent any unnamed cities from getting into the map file.
I think we should at least log an error if we run into this. Thorsten -- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

2012/2/21 Thorsten Kukuk <kukuk@suse.de>:
On Tue, Feb 21, Steve Ratcliffe wrote:
It does include the two example you gave however. Both these are place=* polygons, with place_name instead of name.
That's why I use --name-tag-list=name,place_name meanwhile.
You probably can and should try to use the place_name tag to give the name, but I think that we should also prevent any unnamed cities from getting into the map file.
I think we should at least log an error if we run into this.
Thorsten
Thanks for your advice Thorsten, I shall try these options. I thought that --name-tag-list was used to restrict the language of the tags. I did not know you could use them that way. Kind regards, Eric

2012/2/21 Thorsten Kukuk <kukuk@suse.de>:
That's why I use --name-tag-list=name,place_name meanwhile.
Hi Thorsten, This worked and do not suffer the freezes when listing the nearest cities. However, these formerly unnamed cities are now duplicates of some cities (I verified this matches the coordinates of the previously unnamed city points). In other words, on the map, some cities have two identical points exactly at the same place. Is there a way to avoid this? The best would be indeed to avoid listing the unnamed cities on the map. Thanks, Eric

Hi Eric, On Wed, Feb 22, Eric Fernandez wrote:
2012/2/21 Thorsten Kukuk <kukuk@suse.de>:
That's why I use --name-tag-list=name,place_name meanwhile.
Hi Thorsten,
This worked and do not suffer the freezes when listing the nearest cities. However, these formerly unnamed cities are now duplicates of some cities (I verified this matches the coordinates of the previously unnamed city points). In other words, on the map, some cities have two identical points exactly at the same place. Is there a way to avoid this? The best would be indeed to avoid listing the unnamed cities on the map.
If you change the rule place=city ... to place=city & name=* ... You should get only citys with a name. Yuo need to adjust that of course for the other place=* rules, too. But this sounds like a bug in the OSM data. Could you give me some examples, where you have duplicate POIs? Then I can check if this is an OSM data bug or how to better solve this. Thorsten -- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

2012/2/22 Thorsten Kukuk <kukuk@suse.de>: Hi,
If you change the rule place=city ... to place=city & name=* ...
I am using the default style in mkgmap, so isn't that a bug in this default style's rule? Shouldn't it be corrected in mkgmap?
But this sounds like a bug in the OSM data. Could you give me some examples, where you have duplicate POIs? Then I can check if this is an OSM data bug or how to better solve this.
Thorsten
Hi, Here are some examples close to Oxford: Sandford-on-Thames: N 51 42.620' W001 13.721' Bayworth: N 51 42.522' W001 16.668' Sunningwell: N 51 42.065' W001 17.042' Shippon: N 51 40.785' W001 18.439' Wytham: N 51 46.489' W001 18.696' Hope this helps, Eric

Am 22.02.2012 11:44, schrieb Eric Fernandez:
2012/2/22 Thorsten Kukuk<kukuk@suse.de>:
Hi,
If you change the rule place=city ... to place=city& name=* ... I am using the default style in mkgmap, so isn't that a bug in this default style's rule? Shouldn't it be corrected in mkgmap? For me it isn't a bug. The information there is a village/town/city/... could be very important. Mainly in not well mapped areas. In a village/.... you will find markets, general help, ....
Henning

2012/2/22 aighes <osm@aighes.de>:
Am 22.02.2012 11:44, schrieb Eric Fernandez:
2012/2/22 Thorsten Kukuk<kukuk@suse.de>:
Hi,
If you change the rule place=city ... to place=city& name=* ... I am using the default style in mkgmap, so isn't that a bug in this default style's rule? Shouldn't it be corrected in mkgmap? For me it isn't a bug. The information there is a village/town/city/... could be very important. Mainly in not well mapped areas. In a village/.... you will find markets, general help, ....
Henning
Hi, I understand your point, but the problem here is having unnamed cities created from polygons, which are duplicates of real city/villages. The best way to handle this would be to correct mkgmap to avoid getting these blank duplicates into the map. The intent is not to remove information. Eric

On Wed, Feb 22, Eric Fernandez wrote:
I understand your point, but the problem here is having unnamed cities created from polygons, which are duplicates of real city/villages. The best way to handle this would be to correct mkgmap to avoid getting these blank duplicates into the map. The intent is not to remove information.
That the name is blank is a bug in your (or the default) style: the polygons all have a name tag: "place_name". That's the tag used for a period of time for places in OSM, and we should use that if no name tag is there. The best solution is really to use the option I wrote already here in the thread. The bigger problem is, that the tags for the place node and the place polygon from your examples are even conflicting :( On of the both (polygon or node) needs to be deleted and the other one should be corrected. But since I don't know the area, I cannot say which of the two is the correct one and fix it. Thorsten -- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

Am 22.02.2012 14:32, schrieb Thorsten Kukuk:
The bigger problem is, that the tags for the place node and the place polygon from your examples are even conflicting:(
On of the both (polygon or node) needs to be deleted and the other one should be corrected. But since I don't know the area, I cannot say which of the two is the correct one and fix it. +1
Henning

2012/2/22 Thorsten Kukuk <kukuk@suse.de>:
That the name is blank is a bug in your (or the default) style: the polygons all have a name tag: "place_name". That's the tag used for a period of time for places in OSM, and we should use that if no name tag is there. The best solution is really to use the option I wrote already here in the thread.
The bigger problem is, that the tags for the place node and the place polygon from your examples are even conflicting :(
On of the both (polygon or node) needs to be deleted and the other one should be corrected. But since I don't know the area, I cannot say which of the two is the correct one and fix it.
Thorsten
I just had a look at these places on OSM and indeed, they have both a POI "Village" (or "Hamlet", etc) with the set name (e.g. Cumnor), and around a polygon also called "Village" with no set name. I assume the blank name comes from this unnamed polygon then. Is this what you mention? So does that violate OSM rules for naming areas/villages? Also, you say the style itself has a bug: so is the bug in the map, in the style, or both? Should all "place=XXX" entries in points style file be followed with "& name=*"? Or should we use the option "--name-tag-list=name,place_name" or simply change mkgmap so that it does not add unnamed place_name to the map? Sorry to be daft but this is a bit confusing, considering the possible bug reasons and workarounds. Finally, if the mkgmap default style has a bug, how to report this? Thanks, Eric

Am 22.02.2012 15:59, schrieb Eric Fernandez:
2012/2/22 Thorsten Kukuk<kukuk@suse.de>:
That the name is blank is a bug in your (or the default) style: the polygons all have a name tag: "place_name". That's the tag used for a period of time for places in OSM, and we should use that if no name tag is there. The best solution is really to use the option I wrote already here in the thread.
The bigger problem is, that the tags for the place node and the place polygon from your examples are even conflicting :(
On of the both (polygon or node) needs to be deleted and the other one should be corrected. But since I don't know the area, I cannot say which of the two is the correct one and fix it.
Thorsten
I just had a look at these places on OSM and indeed, they have both a POI "Village" (or "Hamlet", etc) with the set name (e.g. Cumnor), and around a polygon also called "Village" with no set name. I assume the blank name comes from this unnamed polygon then. Is this what you mention? So does that violate OSM rules for naming areas/villages?
Also, you say the style itself has a bug: so is the bug in the map, in the style, or both? Should all "place=XXX" entries in points style file be followed with "& name=*"? Or should we use the option "--name-tag-list=name,place_name" or simply change mkgmap so that it does not add unnamed place_name to the map? Sorry to be daft but this is a bit confusing, considering the possible bug reasons and workarounds. The bug is in osm-data. But with changing the style, you could remove the effects of the bug in osm-data. But I think this shouldn't be done in default-style because of the side-effects.
Finally, if the mkgmap default style has a bug, how to report this? Write a mail to this mailinglist ;)
Hening

2012/2/22 aighes <osm@aighes.de>:
The bug is in osm-data. But with changing the style, you could remove the effects of the bug in osm-data. But I think this shouldn't be done in default-style because of the side-effects.
Finally, if the mkgmap default style has a bug, how to report this? Write a mail to this mailinglist ;)
Hening
:) thanks for your advice. I had a look at the style files. There are "place=" entries in "polygons" style file: place=village [0x03 resolution 19] place=island & name=* [0x53 resolution 19] place=islet & name=* [0x53 resolution 20] and indeed, place=village is not followed with "& name" like the others. Is this normal? Furthermore, there could be other place= where name is unset and is not a village, such as hamlets. Thus as you write it, it would be better to fix this upstream. I'll do a formal post regarding this issue soon. Eric

Am 22.02.2012 16:32, schrieb Eric Fernandez:
2012/2/22 aighes<osm@aighes.de>:
The bug is in osm-data. Do you mean the code that is in ./src/uk/me/parabola/mkgmap/reader/osm of the mkgmap source archive? No, it is in the osm-data which you used for generating your map. If there is a polygon then there shouldn't be a node for the same object.
Henning

2012/2/22 aighes <osm@aighes.de>:
Am 22.02.2012 16:32, schrieb Eric Fernandez:
2012/2/22 aighes<osm@aighes.de>:
The bug is in osm-data. Do you mean the code that is in ./src/uk/me/parabola/mkgmap/reader/osm of the mkgmap source archive? No, it is in the osm-data which you used for generating your map. If there is a polygon then there shouldn't be a node for the same object.
Henning
Ah ok, so the problem is the map itself: a polygon should get the name of the village/hamlet etc and no node should be created on the map. Unfortunately, this seems not be enforced on OSM, considering the number of cases on the GB map. And this is a problem on some Garmin devices. Is there any way to automatically correct this at the level of OSM? At the end of the day, it is the normalisation of the map which is affected there. Incidentally does this mean I should not request a fix for mkgmap on this mailing list? Maybe it would be useful to get a warning, or a supplementary option which filters the unnamed areas? At the end of the day, this only occurs with --add-pois-to-areas, where an area is created but the poi name is blank. Or maybe, could --add-pois-to-area be improved to discard blank names? Eric

Hi, there isn't an error in mkgmap. Your problem is, that thee are 2 osm-objects for one object in reality. You could handle this error different. Eg.: * fix osm-data in osm-db * fix your osm-extract on your pc * change your points-style-file The best would be to fix the error in osm-db. --add-pois-to-area does what it should do. Create for each area a node. There shouldn't be any tests in the code. such tests you can implement in your style-file. Henning

2012/2/22 aighes <osm@aighes.de>:
Hi, there isn't an error in mkgmap. Your problem is, that thee are 2 osm-objects for one object in reality.
You could handle this error different. Eg.: * fix osm-data in osm-db * fix your osm-extract on your pc * change your points-style-file
The best would be to fix the error in osm-db.
--add-pois-to-area does what it should do. Create for each area a node. There shouldn't be any tests in the code. such tests you can implement in your style-file.
Henning
Thanks Henning for the information. You say there shouldn't be any test in the code, but we have some already: --report-undefined-nodes or --remove-short-arcs. Logging level provides messages which are useful, and this issue might be worth logging. Cheers, Eric

On Wed, Feb 22, Eric Fernandez wrote:
2012/2/22 Thorsten Kukuk <kukuk@suse.de>:
That the name is blank is a bug in your (or the default) style: the polygons all have a name tag: "place_name". That's the tag used for a period of time for places in OSM, and we should use that if no name tag is there. The best solution is really to use the option I wrote already here in the thread.
The bigger problem is, that the tags for the place node and the place polygon from your examples are even conflicting :(
On of the both (polygon or node) needs to be deleted and the other one should be corrected. But since I don't know the area, I cannot say which of the two is the correct one and fix it.
Thorsten
I just had a look at these places on OSM and indeed, they have both a POI "Village" (or "Hamlet", etc) with the set name (e.g. Cumnor), and around a polygon also called "Village" with no set name.
Haven't checked Cumnor, but the one I checked have one: place_name They are not without name, we only don't evaluate the alternate tag.
I assume the blank name comes from this unnamed polygon then. Is this what you mention? So does that violate OSM rules for naming areas/villages?
Please read again what I wrote above: place_name is used, no OSM rules are violated.
Also, you say the style itself has a bug: so is the bug in the map, in the style, or both? Should all "place=XXX" entries in points style file be followed with "& name=*"? Or should we use the option "--name-tag-list=name,place_name" or simply change mkgmap so that it does not add unnamed place_name to the map? Sorry to be daft but this is a bit confusing, considering the possible bug reasons and workarounds.
The bug is that place_name is not used for places. It doesn't matter how you fix it, you can fix it in a lot of different ways. Thorsten -- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

This worked and do not suffer the freezes when listing the nearest
There are really four overlapping issues here, so let break them down so they can be dealt with individually. 1. A bad map that causes freezing on some devices. I want to avoid this no matter what the input data is. It seems that avoiding putting unnamed cities into the output file is sufficient in this case, so that is what I think we should do. 2. The add-poi-to-area option used to add a poi only if there was not one there already. I don't think the option does this any more? It think that was a useful feature. It may be more difficult to do now that they are created earlier on however. 3. The default style should be prepared to use place_name for places created from areas 4. The OSM data may be wrong. Well, we mostly have to deal with it as we find it, unless it doesn't match reality. I don't even think that have a poi for a town and an area for the same town is a bad thing. ..Steve

2012/2/24 Steve Ratcliffe <steve@parabola.me.uk>:
This worked and do not suffer the freezes when listing the nearest
There are really four overlapping issues here, so let break them down so they can be dealt with individually.
1. A bad map that causes freezing on some devices.
I want to avoid this no matter what the input data is. It seems that avoiding putting unnamed cities into the output file is sufficient in this case, so that is what I think we should do.
2. The add-poi-to-area option used to add a poi only if there was not one there already.
I don't think the option does this any more? It think that was a useful feature. It may be more difficult to do now that they are created earlier on however.
3. The default style should be prepared to use place_name for places created from areas
4. The OSM data may be wrong.
Well, we mostly have to deal with it as we find it, unless it doesn't match reality. I don't even think that have a poi for a town and an area for the same town is a bad thing.
..Steve _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
Thanks Steve. I think you have accurately summarised the issue. Now I am not familiar enough with mkgmap code and community to suggest the best way to handle this situation. I will leave the decisions to the developers. But I am available to help and eager to test new solutions when required. Cheers, Eric

2012/2/21 Steve Ratcliffe <steve@parabola.me.uk>:
Hi I might have found it, or at least part of the problem. In your map there are 10 cities without a name, created from polygons.
Since there are around 500 cities in the map, this is closer to 1 in 50 than 1 in 10-20 so might not be everything -- do you see more than 10 or is it possible that is the total number of blank lines?
It does include the two example you gave however. Both these are place=* polygons, with place_name instead of name.
You probably can and should try to use the place_name tag to give the name, but I think that we should also prevent any unnamed cities from getting into the map file.
..Steve
Many thanks, I am going to regenerate the map as you suggest. Regarding the frequency, it can indeed be lower than what I reported. I cannot ascertain the total number of cities without a name on the Edge, since the nearest city list is limited in number, and thus the frequency of these blank cities is biased (possibly higher in South Oxford). The best solution would indeed be to prevent unnamed cities to appear in the map file. The fact that the Edge 705 freezes when index is activated is not clear either, but I am pretty sure that without these blank lines there would be no issue. In the source, I saw that the build up of the city list may be related to the MdrBuilder.java file, but am not entirely sure. Kind regards, Eric
participants (5)
-
aighes
-
Eric Fernandez
-
michael lohr
-
Steve Ratcliffe
-
Thorsten Kukuk