I have tested a gmapsupp.img created by mkgmap on my Oregon. It can't find any street when skipping the place name, so it seems you need to specify the place name, otherwise it wont work. This does not happen with the trunk version, it can also find streets if no place name is entered. With a gmapsupp.img created by Mapsource the mixed index branch works ok when skipping the place name.

I tested the branch, using directly the gmapsupp.img file in my device: it is working as expected (lots of tests...) except you can't always search correctly for an address when the same last word is used in different streets names. For example in Padova city (Italy) the following exists: Via Monte Santo Via Del Santo Via Girolamo Dal Santo Piazza Del Santo with the mixed-index, if I type "Santo" I get only: Santo in the list and if I select it shows the address in Via Monte Santo - the others are ignored. In general it should be beneficial to add as a suffix the remaining of the street name when the word is the last one: I find much more user friendly the behaviour I obtained in the main branch with my custom style, where if I type Santo I get: Santo, Via Monte Santo, Via Del Santo, Piazza Del Santo, Via Girolamo Dal in general the last word is always the most significant but it helps to know what was the full name to distinguish different streets named after different persons with same last name (or similar cases); not sure if this can be a general need, for reference below is the style I used with the main branch. Pls. let me know your point, Enrico # Get the last full word if a \s (whitespace) exist in name # if the last full word is a roman number - i.e. if a street has been named after a King or # a Pope - get the last two words # set the labels used for address search (3&4): # the 3rd label is set with the last:word as 1st word followed by comma and the remaining words # the 4th label skipping the 1st word (that is usually Via, Rue, Avenida etc, so not really useful in search) ( mkgmap:country=ITA | mkgmap:country=FRA | mkgmap:country=CHE | mkgmap:country=ESP) & highway=* & name ~ '.*\s.*' { set last:word='${name|part: :-1}' } last:word ~ '(I|II|III|IV|V|VI.*|IX|X|XI.*|XV.*|XX.*)' {set last:word='${name|part: >-3}' } last:word=* { set mkgmap:label:3='${last:word}, ${name|part: <-1}';set mkgmap:label:4='${name|part: >1}' } On Fri, Jan 9, 2015 at 4:12 PM, Enrico Liboni <eliboni@gmail.com> wrote:

How could I have missed this one...! What is exactly the expected behavior? In the gmappsupp.img file each word of the street name should be indexed apart from a country-based list of stop words (pending to be defined) - right?

Currently I workaround the original issue working with just the style regex/substitution/concatenation and the 4 alternate names, it works really well for Italy (at least) since apart form some exceptions that can be handled via proper regexp, the real meaningful word is always the last one or the one after the first (or the full name).

Pls. confirm the expected behavior of this trunk I'll do a test for sure shortly.

Thanks, Enrico

On Fri, Jan 9, 2015 at 12:32 AM, Steve Ratcliffe <steve@parabola.me.uk> wrote:

...

Hi Gerd

my impression was that everybody was happy with the results.

...

What do you think?

Yes I do want to push on and get it merged this time. I think that there may still be a problem with the gmapsupp index and I think we should have a way of dealing with removing common words, especially when they are first as in 'Rue', 'Calle' etc. Doesn't have to be complete, just ready to be used and improved.

I'll write some more tomorrow.

..Steve

_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

Bernd, thanks for the hint - so the style rule can be useful to other non-latin countries, with a proper stop word list to avoid indexing not-meaningful words. In any case - by including a proper list of ignore words per country to limit index size and if the issue I highlighted earlier in this email is solved - the mixed index appears to be the proper final solution to me, without the need to have any additional style rule . Did I get it right that with the mixed index all the words (except if in ignore list) are indexed as single words? Is there a limit in the number of words for a single street? (there might be streets with 6/7 words, using blanks, apostrophes or dashes as separator). On the other side, I like the current output I get with my style rule, where a user can get the street full name also when searching with a single word, that's what I seem to miss with the mixed-index. Enrico On Mon, Jan 12, 2015 at 12:19 PM, Bernd Weigelt <weigelt.bernd@web.de> wrote:

Am Sonntag, 11. Januar 2015, 22:11:43 schrieb Enrico Liboni:

...

Pls. let me know your point,

Hi Enrico

when playing around with your rules, i have seen that in the german speaking parts of Italy, Switzerland. France, Luxembourg and Belgium a lot of streets are tagged with 'An, Zum, In, ...' This parts of the name are also useless for searching by name.

But if i add mkgmap:country=LUX | mkgmap:country=DEU | mkgmap:country=BEL | mkgmap:country=AUT to the country list, i got lots of list entries like Straße, Kieler Weg, Bonner ...

This are my changes

### # Get the last full word if a \s (whitespace) exist in name # if the last full word is a roman number - i.e. if a street has been named after a King or # a Pope - get the last two words # set the labels used for address search (3&4): # the 3rd label is set with the last:word as 1st word followed by comma and the remaining words # the 4th label skipping the 1st word (that is usually Via, Rue, Avenida etc, so not really useful in search)

( mkgmap:country=ITA | mkgmap:country=FRA | mkgmap:country=CHE | mkgmap:country=ESP | mkgmap:country=BEL | mkgmap:country=LUX | mkgmap:country=DEU | mkgmap:country=AUT ) & highway=* & name ~ '.*\s.*' { set last:word='${name|part: :-1}' }

# the ignore list should be greater last:word=* & ( last:word = Straße| last:word = Weg| last:word = Ring| last:word = Platz| last:word = Straat| last:word = Laan ) { delete last:word }

last:word ~ '(I|II|III|IV|V|VI.*|IX|X|XI.*|XV.*|XX.*)' {set last:word='${name|part: >-3}' }

last:word=* { set mkgmap:label:3='${last:word}, ${name|part: <-1}'; set mkgmap:label:4='${name|part: >1}' }

# only for the tests last:word=* {echo '${mkgmap:label:1} | ${mkgmap:label:2} | ${mkgmap:label:3} | \ ${mkgmap:label:4}'; echotags ''} ###

the result is something like that:

320136003: Am Stadtpark | null | Stadtpark, Am | Stadtpark 320136003 - [mkgmap:admin_level6=Rheinisch-Bergischer Kreis,mkgmap:admin_level5=Regierungsbezirk Köln,name=Am Stadtpark,mkgmap:postal_code=42799,mkgmap:postcode=42799,mkgmap:street=Am Stadtpark,route_ref=255,694,255,highway=residential, mkgmap:country=DEU,mkgmap:admin_level2=DEU,last:word=Stadtpark, mkgmap:label:3=Stadtpark, Am ,mkgmap:label:1=Am Stadtpark,mkgmap:admin_level4=Nordrhein- Westfalen,mkgmap:city=Leichlingen,mkgmap:region=Nordrhein- Westfalen,mkgmap:admin_level8=Leichlingen,mkgmap:label:4=Stadtpark ]

-- amarok2 now playing: artist: Lemar title: Don't Give It Up album: Time To Grow

_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev