Opened 10 years ago

Closed 10 years ago

Last modified 10 years ago

#514 closed defect/bug (fixed)

Search and "is_in" is bound to be the wrong solution!

Reported by: flabrosse Owned by: KaZeR
Priority: major Milestone:
Component: core Version: git master
Severity: Keywords: Search
Cc: fred@…

Description

Hello all,

I'm a recent user of navit but so far haven't been able to use it for real. One of the reasons is the search. I read on the wiki that only cities with a "is_in" field can be searched for and that you only want a few per country so that the search is not too slow. So I little by little add this field to places I'm searching for with values of near cities for which the search worked.

However, looking at more and more of these, I found such an incredible variety of is_in fields that this is doomed to fail. For example, there is in France a village called Garancieres-en-Beauce that navit finds and a few kiloemetres away from it another one called Saint-Escobille, which Navit can't find. The reason seems to be that the former has "Europe,France,Eure-et-Loir" for its is_in and the later has "Europe,France,Essonne" for its is_in. I couldn't find any in Essonne that could be found by navit.

Now given the variety in the content of the is_in field (apparently per departement in France, of which there are around 100, less formal in the UK), you are bound to need more than 3 or 4 is_in values per country.

Does all this make sense, or did I completely miss the point?

Cheers,

Fred

Change History (4)

comment:1 in reply to: ↑ description ; follow-up: Changed 10 years ago by cp15

Replying to flabrosse:

Now given the variety in the content of the is_in field (apparently per departement in France, of which there are around 100, less formal in the UK), you are bound to need more than 3 or 4 is_in values per country.

It should be sufficient if France is listed in the is_in tag. The position doesn't matter, and the other names also. Can you give me the osm ids of the two examples you mentioned or at least the coordinates to see what wents wrong?

We are planning to use checking against the country boundary soon, but a fast point in polygon test for this still needs implementation.

Bye,

CP15

comment:2 in reply to: ↑ 1 Changed 10 years ago by flabrosse

Replying to cp15:

Replying to flabrosse:

Now given the variety in the content of the is_in field (apparently per departement in France, of which there are around 100, less formal in the UK), you are bound to need more than 3 or 4 is_in values per country.

It should be sufficient if France is listed in the is_in tag. The position doesn't matter, and the other names also. Can you give me the osm ids of the two examples you mentioned or at least the coordinates to see what wents wrong?

We are planning to use checking against the country boundary soon, but a fast point in polygon test for this still needs implementation.

So it's not the entire tag that counts, only words of it. That is good to know.

I don't know how you get the osm id but the lat,long of the two vilages are:

Garanciere (one that works): lat 48.4357, long 1.918 Saint-escobille (one that does not work): lat 48.43, long 1.9658

Cheers,

Fred

comment:3 Changed 10 years ago by flabrosse

  • Resolution set to fixed
  • Status changed from new to closed

I just installed svn revision 2849 on my Nokia N800 and the search now works... I don't know if anybody did anything on it since 22h ago when I reported the problem, but this is great!

Plus what cp15 said about the search probably means that this is not a problem any more, or at least will not be applicable soon. I will therefore change the status to fixed.

Cheers,

Fred

comment:4 Changed 10 years ago by flabrosse

I just installed today's svn revision on my desktop machine, where I was doing the tests that I reported in the original ticket and it doesn't work. I then I realised that this is because I was using an old map... I have now updated the map and it all works.

I can only apologise for the trouble. Sorry.

Fred

Note: See TracTickets for help on using tickets.