[mkgmap-dev] r3678 and overview map problems

Sat Jul 16 14:07:42 BST 2016

Hi Thomas,

I thought about this a bit more and I think the patch is too complex and on the other hand does not catch all possible cases.

The real problem here is that we cannot store the tile size (0x4a) info for a rather large tile in an overview

map (ov) that has a rather high resolution , e.g. 20.

I now think mkgmap should 1st find the largest tile that has to be stored in the overview map,

the dimensions of this tile determine the highest possible resolution.

This should be done even when ovm_*.img files are found and those tiles contain conflicting

resolution levels (taken from the overview-levels option).

I've coded this different approach in the attached patch, a new binary is here:
http://files.mkgmap.org.uk/download/300/mkgmap.jar

This patch no longer requires an extra read of the input files nor the overview-levels option,
so I think it is the better solution

Gerd

________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Thomas Morgenstern <webmaster at img2ms.de>
Gesendet: Freitag, 15. Juli 2016 19:01:32
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] r3678 and overview map problems

I have tested a few more mapset's, I found no problem. thank you

Thomas

Am 15.07.2016 um 07:34 schrieb Gerd Petermann:
> Hi all,
>
> I assume Thomas did not find any problems. I'd like to commit the patch.
> Any complains?
>
> Gerd
>
>
> Thomas Morgenstern wrote
>> Hi Gerd, thank's for the solution. I have testet the patched r-3678.jar
>> binary for two examples. It works  well.  But i must make a few more
>> test's. Till now, all is okay..
>>    Thomas
>> Am 09.07.2016 um 08:14 schrieb Gerd Petermann:
>>> Hi Thomas,
>>>
>>>
>>> okay, I have now checked your test case as well.
>>>
>>> I think Steve has already explained why the change in r3674 introduced
>>> the problem,
>>>
>>> the newer versions read the input files in sorted order when creating
>>> the overview map,
>>>
>>> older releases used the order given in the command line.
>>>
>>>
>>> A simple work-around would be to use higher numbers for the tiles
>>> which should be processed later,
>>>
>>> e.g. 9999 instead of 4000 for the contour tiles.
>>>
>>>
>>> A better solution would be to detect the needed resolution. The
>>> problem here:
>>>
>>> mkgmap has to read all input tiles (completely) to find out which one
>>> uses the lowest resolution,
>>>
>>> current code doesn't allow to read only the levels info. So one has to
>>> accept a longer run time.
>>>
>>> I don't know if that is really needed, maybe an empty overview map can
>>> always use a low resoltion
>>>
>>> like 12?
>>>
>>>
>>> Another solution would be to evaluate the overview-levels option as
>>> suggested by Steve.
>>>
>>>
>>> In your case the overview map is empty, it contains only the 0x4a
>>> polygons, so it is quite simple.
>>>
>>>
>>> I have no idea what mkgmap should do when a user tries to combine
>>> ovm*.img files which were
>>>
>>> created with different overview-level options, I leave that open.
>>>
>>>
>>> I've impemented both alternatives in the attached patch, a binary
>>> based on r3678 is here:
>>>
>>> http://files.mkgmap.org.uk/download/299/mkgmap.jar
>>>
>>>
>>> The patch solves your problem, but I'd prefer to avoid the additional
>>> reading of all input files
>>>
>>> before the overview map is produced.
>>>
>>> @all : What do you think about the alternative to use a fixed
>>> resolution of 12 for an overview map
>>>
>>> that contains only the tile boundary infos?
>>>
>>>
>>> Gerd
>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>> *Von:* mkgmap-dev <
>> mkgmap-dev-bounces at .org
>> > im Auftrag
>>> von Thomas Morgenstern <
>> webmaster@
>> >
>>> *Gesendet:* Freitag, 8. Juli 2016 20:41:08
>>> *An:* Development list for mkgmap
>>> *Betreff:* Re: [mkgmap-dev] r3678 and overview map problems
>>> Hi Gerd , Update:
>>>
>>> r-3778 create now 1 0x4a like expected. This is a good news. But in my
>>> testenvironment are 2 osm-tiles and 1 contourtile. One of the osmtile
>>> has a wrong position. it is shifted to east ~22 degrees. The tile
>>> itself has 25.4 degreees east-west .
>>> thomas
>>>
>>> Am 08.07.2016 um 16:49 schrieb Thomas Morgenstern:
>
>
>
>
> --
> View this message in context: http://gis.19327.n5.nabble.com/r3678-and-overview-map-problems-tp5877609p5878424.html
> Sent from the Mkgmap Development mailing list archive at Nabble.com.
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
>
>

_______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20160716/fb6b216e/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: overview_levels_v2.patch
Type: application/x-download
Size: 4572 bytes
Desc: overview_levels_v2.patch
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20160716/fb6b216e/attachment.bin>