nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Which values are selected for cross-tab?

h3llraid3r

Hi,
I run into some understanding problems:

I've got data-sets which have 5 columns:

- server name
- server group
- value
- date as string (no possibility to change that, because data-source is a flat file)
- date as date time (computed colum)

Based on this data-sets, I created data-cubes:

- server name as dimension
- server group as dimension
- date as dimension (year, month, day of month, hour selected as "sub-dimensions")
- value as dimension
- value es summary field (MAX as function)

When I create a cross-tab using these dimensions/summaries, there seems to be a problem:
I selected date-time(hour) as row definition, server name as column definition and the value(summary field) as summarize field.

Selecting measure["COLUMN_VALUE"] and aggregation on hour and server name, the MAX-function returns the right value (maximum value for each server in every hour which is displayed).

But when I use the AVG-function, the values which are displayed, are definitly wrong.

This lead me to a understanding problem, which values are selected by the data-cube respectivly by the cross-tab? I thought, the values were selected correctly, because of the right MAX-value for each server. If the values would be wrong, these values would be wrong, too?!

Find more posts tagged with

Comments

JasonW

Can you post the report?

Jason

h3llraid3r

For a better understanding, I attached some pictures, which may help to understand my problem.

Origin data from server A:

h3llraid3r

Ok, problem solution was very easy...

After adding the original time column (which is imported from flat file as text and computed to a date-formatted column) into the data cube, the values are correct.

I don't understand what that changes for BIRT (sorting/filtering for an text-formatted date would not be logical), but I can live with that

If someone knows about that issue, an explanation would be very kind for a better understanding in BIRT.

Greets,

h3llraid3r

JasonW

In the original flat file dataset did you set the column type to a datetime?

Jason

h3llraid3r

The flat file is a *.csv-file, so there was no option to directly set the column type to a date-time. So I created a computed column "time", which is date-type and based on the data of the time-column.

I got it working now (see above), but I still don't understand why BIRT must have the text-formatted date-column in the data cube. The computed time-column (an that's what I'm working with in the report) naturally was in the data-cube, but without adding the original column to the data-cube, that didn't work.

JasonW

You may want to log a bug for this, but in the flat file data source in the select columns node you should be able to set the datatype to datetime instead of string(which is the default).

Jason

h3llraid3r

As you can see here,

JasonW

Click on the select columns node and if you click on the datatype there you should be able to change it.

Jason

h3llraid3r

Ok, I haven't seen that yet... Would have made things a bit easier