[mkgmap-dev] MDR and city streets on different tile from POI place=city

Thu Nov 4 09:55:28 GMT 2021

Hi Ticker,

I fear you lost me. My current understanding is that it doesn't matter much if there is one city or two as long as the index structures are correct and Garmin software finds the objects that exist. MapSource seems to ignore the case differences when it builds the list of city names in the drop down lists, it only showed one entry in my tests. I've not yet tried to understand which variant is choosen, might be the first in the MDR5  section. Anyway, it presents results with different spelling and they all seem to be correct as the correct object is shown when I click on an entry.

Just post the minimal patch(es) and describe the expected effect(s) .

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Ticker Berkin <rwb-mkgmap at jagit.co.uk>
Gesendet: Mittwoch, 3. November 2021 17:47
An: Development list for mkgmap
Betreff: Re: [mkgmap-dev] MDR and city streets on different tile from POI place=city

Hi Gerd

As per your example with "De Wijk", it could be the same city and
different cities and difficult to decide programatically - esp. if
different tiles.

My attempts to cope with this in the MDR indexes didn't seem to work
and I've abandoned this idea. I'd like to release a minimal patch that
fixes the 2 assertion crashes. With --lower-case, which streets are
attached to the different case variants of a city will not bear
scrutiny - I don't think it did before.

Then we should considering case-sensitivity of street, city, region and
country as a new issue.

Ticker

On Wed, 2021-11-03 at 14:27 +0000, Gerd Petermann wrote:
> Hi Ticker,
>
> yes, might be better to change that to be more precise. Question is
> if Aa and AA are two different cities or just different spellings of
> the same city. The latter is much more likely, so it seems justified
> to use only one.
> We have more code which tries to ignore case, e.g. for --housenumbers
> I often use String.equalsIgnoreCase() to work around different
> spelling in road names and addr:street or place names and addr:place.
>
> Gerd
>
> ________________________________________
> Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag
> von Ticker Berkin <rwb-mkgmap at jagit.co.uk>
> Gesendet: Mittwoch, 3. November 2021 11:40
> An: Development list for mkgmap
> Betreff: Re: [mkgmap-dev] MDR and city streets on different tile from
> POI place=city
>
> Hi Gerd
>
> The scenario is:
>  --lower-case,
>  2 tiles,
>  mkmap:city "Aa" area that spans the tiles",
>  streets with this mkgmap:city in either/both tiles,
>  the City POI "Aa" on one of the tiles,
>  another city, eg "AA" in the tile without the "Aa" POI,
>  streets for this.
>
> I am finding streets in the tile with both regions are all assigned
> to
> one or the other and this behaviour is effected by the presence of
> the
> City POI.
>
> PlacesFile::CreateCity(not unique) will return the same City record
> for
> both "Aa" and "AA" so I propose to remove the toUpperCase() from it
> when --lower-case (unless ascii/cp0 which forces upper case)
>
> Ticker
>
>
> On Wed, 2021-11-03 at 09:17 +0000, Gerd Petermann wrote:
> > Hi Ticker,
> >
> > no idea what your problem is.
> > Address search seems to work fine without any city POI. I think the
> > code just tries to avoid duplicated entries in the LBL file?
> >
> > Gerd
> >
> > ________________________________________
> > Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag
> > von Ticker Berkin <rwb-mkgmap at jagit.co.uk>
> > Gesendet: Dienstag, 2. November 2021 19:31
> > An: Development list for mkgmap
> > Betreff: Re: [mkgmap-dev] MDR and city streets on different tile
> > from
> > POI place=city
> >
> > Hi Gerd
> >
> > Actually, a major part of my problem is probably app/lbl/PlacesFile
> > with its toUpperCase()
> >
> > Ticker
> >
> > On Tue, 2021-11-02 at 16:52 +0000, Ticker Berkin wrote:
> > > Hi Gerd
> > >
> > > I've been checking the street list in the Mdr20 index to make
> > > sure
> > > they
> > > are attached to the correct city (as part of checking case-
> > > significance
> > > for the unicode/mdr errors) and finding discrepancies.
> > >
> > > I think this is because combiners/MdrBuilder, on a tile by tile
> > > basis
> > > takes the POI city list as the basis for Mdr5Records for
> > > attaching
> > > to
> > > streets and passing on for MDR processing.
> > >
> > > If the POI is in an adjacent tile to a road with the same
> > > mkgmap:city,
> > > there seem to be problems.
> > >
> > > This might be the reason for not being able to find some streets
> > > by
> > > city, rather than problems with case.
> > >
> > > I'm still trying to work it all out.
> > >
> > > Ticker
> > >
> > >
> > > _______________________________________________
> > > mkgmap-dev mailing list
> > > mkgmap-dev at lists.mkgmap.org.uk
> > > https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> >
> >
> > _______________________________________________
> > mkgmap-dev mailing list
> > mkgmap-dev at lists.mkgmap.org.uk
> > https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> > _______________________________________________
> > mkgmap-dev mailing list
> > mkgmap-dev at lists.mkgmap.org.uk
> > https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>
>
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

_______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev