Creating QVD ends in loss of records

[Anonymous]

Hi everybody,

I'm trying to use QVD for a large amount of records.

I got the problem, that not every original record was written into my new QVD.

The only difference from my old data model is that I now swap out my data

previously in QVD instead to use them directly in my tables and calculations.

I used the following commands:

table1_qvd:

LOAD    *

FROM

table1.csv

(txt, codepage is 1252, embedded labels, delimiter is ';', msq);

store table1_qvd into ;

...

After that I use the Resident-statement in my tables.

I should perhaps add that I only testweise get my data from csv files,

because everything after that is to come from an SQL database.

Does anyone have a solution? Thank You!

[marcus_sommer]

How did you load the qvd in excel - creating and exporting a tablebox in qlikview?

[Anonymous]

Yes.

[marcus_sommer]

This didn't worked because a tablebox ignored duplicates. To be sure that a tablebox showed each record you need to add an unique record-id within the tablebox mostly generated per recno() or rowno() like above suggested.

- Marcus

[Anonymous]

There should be no dublicates in my csv. They were generated as distinct.

In the graph of the data model, the number of records is indeed displayed,

which is far too low.

[marcus_sommer]

If all of your records are unique have you used all fields within the tablebox? An example, if you have a boolean field as any flag within your table and you put only this flag-field in the tablebox you will be get only to records one with true and one with false - it's the same like a listbox. Therefore are you sure that within the tablebox-content are not duplicates?

- Marcus

[swuehl]

If you are sure that the QVD shows less records than the csv file, could you elaborate this a bit more detailed?

Can you describe what steps you have followed to make the comparison?

Can you give some numbers? (How many records expected / received? QVD stopped at certain records or are there intermittent records missing?)

Are you really sure that writing the QVD is the problem, and not already reading in the csv file (for example, QV not being able to fully interprete all delimiters)?

Could you post the document log of your script, stripped down to a minimum (would be even better if you could upload your csv, but I assume that's not feasible?), like

table1_qvd:

LOAD    *

FROM

table1.csv

(txt, codepage is 1252, embedded labels, delimiter is ';', msq);

store table1_qvd into ;

DROP TABLE table1_qvd;

LOAD * FROM (qvd);