Format of Column Headers in database resulting from Database Writer node



KNIME is a wonderful tool and I am making very good progress on my task.  My congratulations to the folks who designed and built this!


I have what I hope is a simple question.  Using the Database Writer node, I am attempting to output my data into a table in my database.  The output works, but the column names are renamed in a way that will confuse joins at a later date, since these same columns were previously read from the same database.  In particular, a table with the column name "This & That" (including the quotation marks) inside KNIME shows up in the database as _This___That_ (with no quotation marks or ampersand, but with five underscore characters). 


Generally I do not know ahead of time what these column names will be, so renaming them by hand (for instance, to add more quotes or backslashes or some other escape character) is not really possible.  Is there a way I can output the column names with their quotation marks?


Many thanks,

Bill N

I can confirm that we do a replacement on all special chars before we write data into the database. This is obviously too restrictive in some cases, but also keeps our nodes general in the sense that they work with most JDBC databases. I open a feature request to switch this behavior on/off in the node dialog. There is currently no work around, except changing the database structure, for it.

I understand.  Many thanks for the feature request.  Might I also ask that a feature request to make the 8 or so parameters in the J48 Weka classifier node changeable by flow variables (and hence by the command line) might also be a very useful extension?  I suppose this might be true for classifiers in general.


Much thanks,

Bill N

Unfortunately not, more about weka and flow variable are available here.