Thread: GPS "New Address" Feedback

I use my carputer for over 1 year now, with different frontends. I have Centrafuse since 3 months now, and i'm pretty sure i will stay with this.
Navigation is used several times a day, but i never navigate using ZIP codes. Personally i think it won't be any loss.

I have managed to load all the postcode / zip files within a couple of seconds.. well in a fashion.

ok.. this is in short how I do it..

First get the count of the list.. ie. zipcount.

If this count is over say 5000 items.. I then split this array up.. ie..

Code:

If ca.ItemCount > 4999 Then
            splitlist = True
            splitby = 1000
            For n = 0 To arraycount - 1 Step splitby
                ReDim Preserve splitlistarray(c + 1)
                If position = 1 Then
                    If searchby = "city" Then splitlistarray(c) = FDest.GetCityNameFromID(ca.Item(n))
                    If searchby = "street" Then splitlistarray(c) = FDest.GetStreetNameFromID(ca.Item(n))
                    If searchby = "postcode" Then splitlistarray(c) = FDest.GetZipCodeFromID(ca.Item(n))
                End If
                If position = 2 Then
                    If searchby = "city" Then splitlistarray(c) = FDest.GetStreetNameFromID(ca.Item(n))
                    If searchby = "street" Then splitlistarray(c) = FDest.GetCityNameFromID(ca.Item(n))
                    If searchby = "postcode" Then splitlistarray(c) = FDest.GetStreetNameFromID(ca.Item(n))
                End If

                If n > ProgressBar1.Value + pb Then ProgressBar1.Value = n : ProgressBar1.Refresh()
                c = c + 1
            Next
            ctto = 1000
        Else
            ctto = ca.ItemCount
            splitlist = False
        End If

This then loads a new array up.. and at its greatest here in the uk 1.2 million postcodes.. it then only holds an array of 1000...

Then Say you want to start searching for a postcode.. you input S..

you then compare each value in the new array. Its then very easy to find which one contains the S your after, you then load just this array group and then search this array group for the needed one.

Using this way proves very fast..

Although the main downfall of using your own address list search is the fact the Destnators list is not in alpha order.. theres lots of pit falls, and I have come to thinking the only real way to do it is to embed the default address search dialog again.

hope this helps, if you want to check out the speed.. download a quick avi from www.freedrive.co.uk/address.zip

CdR

p.s. also the avi taken is playing in the IDE, so its also faster than you see when its compiled ;-)

Originally Posted by CdRsKuLL

I have managed to load all the postcode / zip files within a couple of seconds.. well in a fashion.

ok.. this is in short how I do it..

First get the count of the list.. ie. zipcount.

If this count is over say 5000 items.. I then split this array up.. ie..

Code:

If ca.ItemCount > 4999 Then
            splitlist = True
            splitby = 1000
            For n = 0 To arraycount - 1 Step splitby
                ReDim Preserve splitlistarray(c + 1)
                If position = 1 Then
                    If searchby = "city" Then splitlistarray(c) = FDest.GetCityNameFromID(ca.Item(n))
                    If searchby = "street" Then splitlistarray(c) = FDest.GetStreetNameFromID(ca.Item(n))
                    If searchby = "postcode" Then splitlistarray(c) = FDest.GetZipCodeFromID(ca.Item(n))
                End If
                If position = 2 Then
                    If searchby = "city" Then splitlistarray(c) = FDest.GetStreetNameFromID(ca.Item(n))
                    If searchby = "street" Then splitlistarray(c) = FDest.GetCityNameFromID(ca.Item(n))
                    If searchby = "postcode" Then splitlistarray(c) = FDest.GetStreetNameFromID(ca.Item(n))
                End If

                If n > ProgressBar1.Value + pb Then ProgressBar1.Value = n : ProgressBar1.Refresh()
                c = c + 1
            Next
            ctto = 1000
        Else
            ctto = ca.ItemCount
            splitlist = False
        End If

This then loads a new array up.. and at its greatest here in the uk 1.2 million postcodes.. it then only holds an array of 1000...

Then Say you want to start searching for a postcode.. you input S..

you then compare each value in the new array. Its then very easy to find which one contains the S your after, you then load just this array group and then search this array group for the needed one.

Using this way proves very fast..

Although the main downfall of using your own address list search is the fact the Destnators list is not in alpha order.. theres lots of pit falls, and I have come to thinking the only real way to do it is to embed the default address search dialog again.

hope this helps, if you want to check out the speed.. download a quick avi from www.freedrive.co.uk/address.zip

CdR

p.s. also the avi taken is playing in the IDE, so its also faster than you see when its compiled ;-)

That is the problem, the are not in alpha order, therefore you can't binarysearch on them and ensure you will get results, therefore you have to sort after you get the array back and that is what is taking the time...

david

I find myself using the zip code more then city only because it shortens the list down on the street step. I still have problems with entering some addresses. E St. Pat for example. Sometimes it doesn't find it, I so I have to manually find it in the list.. So the zip narrows it down so I am not looking at all 3 or 4 zip code worth of streets of a city...

It would be nice if the map would remember the last city/zip selected. I don't normally select a different city each time I enter a new address mainly the street.

Originally Posted by veetid

That is the problem, the are not in alpha order, therefore you can't binarysearch on them and ensure you will get results, therefore you have to sort after you get the array back and that is what is taking the time...

Like I mentioned in the bugs thread, since the data is static, you can pre-sort them, write it all to a binary file that includes a pointer to the rest of the data that you key off of the zipcode for.

There's no need to have to sort that data on-the-fly all the time.

Originally Posted by midiwall

Like I mentioned in the bugs thread, since the data is static, you can pre-sort them, write it all to a binary file that includes a pointer to the rest of the data that you key off of the zipcode for.

There's no need to have to sort that data on-the-fly all the time.

This is what I have done... It defaults to CITY/STREET search when you click new address, but if you hit back it gives you the list of new address options...

The first time you load street or zip searching it takes a few seconds, but then stores the information so it loads instantly the next time you use it...

This takes some memory, but works on my M10000...

david

Originally Posted by veetid

This is what I have done... It defaults to CITY/STREET search when you click new address, but if you hit back it gives you the list of new address options...

The first time you load street or zip searching it takes a few seconds, but then stores the information so it loads instantly the next time you use it...

This takes some memory, but works on my M10000...

Cool! I'd say that a pre-load time is very acceptable, 'specially considering that most folks are using Hibernate, so the pre-load is only really when the machine boots.

Well done!