February 10, 2012 at 3:43 am
I get warnings in my package because I have some unused columns coming from a Union All task.
The "Union All" task gets all its columns from the first input and does not allow me to choose the columns on output.
I suppose its just a continuation of the input buffer so how could I "remove" a column?
Am I missing something (again)?
February 10, 2012 at 3:57 am
Would you not just remove a column from the input?
John
February 10, 2012 at 4:23 am
Actually, it complains about three columns.
One of these columns is not really used anywhere indeed so I removed it from the derived column task where I created it.
The bizarre thing is that it was much higher in the dataflow so why complain on the "Union All" task only?...
For the two other columns, I still don't understand because they are both used in a "OLE DB Command" task just feeding the "Union All" task.
This time, I can't remove these columns earlier since i need them???
...
Anyway, we're making progress: 1 warning out of 3 down! 🙂
February 10, 2012 at 4:31 am
Can you show us what your data flow looks like, please?
Thanks
John
February 10, 2012 at 5:00 am
In the attached screenshot,
1) is where I use these variables
2) is where I get the warning
February 10, 2012 at 6:57 am
OK, so the MAJ Clients OLE DB Command only feeds the Union All (Tous les clients) transformation? Can you not restrict the number of columns in the output of MAJ Clients?
John
February 10, 2012 at 7:33 am
Eric Mamet (2/10/2012)
I get warnings in my package because I have some unused columns coming from a Union All task.The "Union All" task gets all its columns from the first input and does not allow me to choose the columns on output.
I suppose its just a continuation of the input buffer so how could I "remove" a column?
Am I missing something (again)?
Right-click the row in the UNION ALL editor and choose Delete.
The column is now removed from the output.
The reason BIDS started complaining at the UNION ALL is probably because the UNION ALL is an asynchronous component, in other words, it recreates the pipeline buffer.
Need an answer? No, you need a question
My blog at https://sqlkover.com.
MCSE Business Intelligence - Microsoft Data Platform MVP
February 10, 2012 at 7:49 am
Damn again, I was looking on right click but I had my cursor in the wrong field and/or the column state was wrong.
Cool!
February 10, 2012 at 7:51 am
I looked at removing the columns from the OLE DB Cmd output but could not find a way.
And is it not impossible to remove a column from a flow?
February 10, 2012 at 8:15 am
If you look at the properties of the Union All editor, you'll get a list of all columns in one input alongside a list of all columns in the other. You can choose here which columns to map to which, and which to ignore. Presumably, the ignored ones will show as NULLs in the output. Your diagram doesn't show what the destination of your output is, but suppose it's an OLEDB Destination. You can then ignore the columns from the input that you don't want. Does that help?
John
February 10, 2012 at 8:18 am
As suggested by Koen, I deleted the columns within the "Union All" by right-clicking on the output column definitions.
February 10, 2012 at 9:26 am
I tried that but the task complains if you don't map any output to an input.
On the other hand, as suggested by Koen, we can right click on a column mapping definition and click the Delete option.
I did not see it initially because I right clicked in the first column (the output) and you don't see the Delete option if you are attempting to edit that column name... Bad luck!
Thanks for your help
Viewing 12 posts - 1 through 12 (of 12 total)
You must be logged in to reply to this topic. Login to reply