Number precision problem with new XLS Reader (Knime 3.2 to 3.3)

Hi dear community,

I recently upgraded to Knime 3.3.1 and I am having a problem with the new XLS Reader.

With the previous version (Knime 3.2.x) of the XLS Reader node, the numbers in spreadsheet are read with their full precision regardless of the cell format.

As an example:

If an excel cell contains the number "3.141593" but the format only displays "3.142", the previous reader extracts the "3.141593" value whereas with the new reader, the value extracted is "3.142".

As the previous node is still executable (but not available in the node repository) I'm still able to get full precision numbers but I am wondering if this problem will be fixed for the new reader node.

Does someone have any information about this problem ?

Thanks !

Hey,

for me, the Excel Reader node works as expected. I got full precision. In your node's output table you can right-click on your column header and choose "Full Precision" as your renderer. Did you know that?

Best,
Marc

Hi Marc,

Thanks for your answer.

I didn't know the tip you mentioned but it still doesn't help as I would have to manually set dozens of columns to full precision which is not a realistic solution.

With the new xls reader, when your columns have only numeric contents, full precision numbers are extracted but when your columns have mixed contents, formatted numbers are extracted instead of full precision ones.

For, me there really should be an option for using or not the formatted values when reading an excel file.

Here is my detailed situation :

Excel displayed values
1950 1951
1.23 NO
NA 0.01
9.88 NA
Excel stored values
1950 1951
1.23456 NO
NA 0.0078
9.87654 NA

When I use the new XLS Reader I got :

Table with new XLS Reader
1950 1951
1.23 NO
NA 0.01
9.88 NA

When I use the old Excel Reader I got :

Table with old XLS Reader
1950 1951
1.23456 NO
NA 0.0078
9.87654 NA

 

Hi Marc,

Thanks for your answer.

I didn't know the tip you mentioned but it still doesn't help as I would have to manually set dozens of columns to full precision which is not a realistic solution.

With the new Excel reader, when your columns have only numeric contents, full precision numbers are extracted but when your columns have mixed contents, formatted numbers are extracted instead of full precision ones.

For, me there really should be an option for using or not the formatted values when reading an excel file.

Here is my detailed situation :

Excel displayed values
1950 1951
1.23 NO
NA 0.01
9.88 NA
Excel stored values
1950 1951
1.23456 NO
NA 0.0078
9.87654 NA

When I use the new XLS Reader I got :

Table with new Excel Reader
1950 1951
1.23 NO
NA 0.01
9.88 NA

When I use the old Excel Reader I got :

Table with old Excel Reader
1950 1951
1.23456 NO
NA 0.0078
9.87654 NA
 

To clarify this:

If you have full numeric columns, they are extracted with full precision. Everything you do with the column, happens with full precision. Regarding my tip from my previous post, this is only how the numbers are depicted in the table view.

Indeed, with mixed content type, there is even not such option.