SQLite Query for populating combobox

[FrancescoDiMuro]

Good evening guys

I am working on a little project, in which I have to retrieve 8000+ rows of data from a table, from a SQLite database, to populate a combobox 

This is what I tried 'til now, but it still takes about 12 seconds to populate the combobox.

_SQLite_Exec($objDatabase, "BEGIN TRANSACTION;")
If _SQLite_GetTable($objDatabase, "SELECT DISTINCT Comune FROM LISTA_COMUNI;", $arrRisultatoQuery, $intRighe, $intColonne) = $SQLITE_OK Then
    For $i = 2 To UBound($arrRisultatoQuery) - 1
        If $i < UBound($arrRisultatoQuery) - 1 Then
            GUICtrlSetData($cbo_ComuneNascita, $arrRisultatoQuery[$i] & "|")
        Else
            GUICtrlSetData($cbo_ComuneNascita, $arrRisultatoQuery[$i])
        EndIf
    Next
_SQLite_Exec($objDatabase, "COMMIT;")

Are there any other solution to retrieve 8000+ records from a SQLite database?

Thank you very much  

[Danyfirex]

Hello. Use _GUICtrlComboBox_AddString is faster than GUICtrlSetData

 

Saludos

[jchd]

@FrancescoDiMuro,

Remove BEGIN and COMMIT. You don't need to enclose a single select (or any other single SQL statement BTW) in a transaction. SQLite will do that for you under the hood.

Now why is selecting distinct values so slow? Do you have duplicates in your table? If yes you should remove extra copies. If you don't have duplicate rows but only several rows for the same "Comune" column, then the design is flawed: you shouldn't duplicate data like that. Use a separate table and foreign keys for that.

To make you understand what I mean, imagine you have a table of all streets in a country. A bad design would be:
Id, Region, City, Street
where Region and City will be repeated for every street.

A better design will use a Regions table, a Cities Table and a Streets table.

Regions:
RegionId, RegionName (Unique)

Cities:
CityId, CityName (Unique in a given region)

Streets:
Id, RegionId, CityId, StreetName
where RegionId is a foreign key to the Regions table and CityId a foreign key to the Cities table.

Of course this is just a simplified example to fix ideas. This process of avoiding data duplication is called DB normalization.

Besides these considerations, you can also use the power of SQL to return you a single string with all the data and delimiters in place, for instance:

_SQLite_QuerySingleRow($objDatabase, "select group_concat(Comune, '|') FROM (select distinct comune from LISTA_COMUNI) group by null;", $Row)

On my table of 75377 zip codes for 32 countries, this request takes 300ms on my very slow PC. This way you can feed $Row to the combobox all in once. Yet, I personally suspect that a combobox with 8000+ entries is definitely unpractical, but that is up to you.

Edited May 26, 2017 by jchd

[FrancescoDiMuro]

Good morning @jchd

The table I have does not contain more than a couple of duplicated rows...

What I'd like to have, is retrieve all the data from "Comune", and display them in a combobox, in order to let the user select what he wants to.

Do you have any pratical suggestion on how to display so many rows, and let select just one of them, in order to manage the data selected?

For SQLite: I can try with what you did suggest to me... But then, I have to split data, am I right?

Thank you a lot  

[jchd]

I find it irritating when I'm offered a choice in some overpopulated, endless and boring listbox of combobox.

Restricting the choices to the few which loosely match user input seems preferable. I've made available an SQLite extension DLL named "Unifuzz" which offers a number of Unicode string and collations functions, among which a typos() function returning the Damerau-Levenshtein distance between two strings. I often use it this way, loosely adapted to what I understand of your context:

select distinct comune from comuni where typos(comune, $input) < 3 order by comune;

This returns all distinct comune that are 4 character change away from the string in variable $input. For instance, searching cities in Belgium,
select distinct ville from codespostaux where paysiso = 'BE' and typos(ville, 'böan') < 3 order by ville;
returns:

Ville
BAAL
BOHAN
BOOM
BOST
BRA
BRAS
BRAY
DOHAN
MEAN
MOEN
ON

You can see that allowing as little as two typos already grabs a fair but manageable number of plausible choices. This function works by first folding case and accents on the whole BMP unicode range. Sorry I don't have a list of italian comuni to examplify.

[Joboy2k]

GUICtrlSetData($cbo_ComuneNascita, _ArraytoString($arrRisultatoQuery, "|"))

 

 

[jchd]

_ArrayToString is a possibility but using the power of fast SQL function will always reveal faster than UDF code.