Topic: Changing data types in .csv, text inputs
Changing data types in .csv, text inputs
When I load data samples from CSV or other text files, the column data types are always set to VARCHAR, and I am unable to perform numeric analysis (profiling) from the interface. I would like to be able to edit the metadata definition.
Actually the option is already there at an API level but not in the user interface.
It's not currently in the plans for DC 1.5 but on the other hand it's not such a biggie so if anyone wants to have a go I wont stop them. Otherwise I will implement it for 1.6.
It's not currently in the plans for DC 1.5 but on the other hand it's not such a biggie so if anyone wants to have a go I wont stop them. Otherwise I will implement it for 1.6.
I reviewed the interface and I can see where a change might be made, but I also ended up with a cascade of requirements that probably require a high-level vision of where the project should go;
(1)load/save metadata defs
(2)load/save the data selection
(3)unify/align .dcp and.dcv files, and probably some other files I have not seen.
I kind of see a a final goal of DataCleaner to capture either an ETL process or a 'fixit-bot,' and my particular need seems to be a distraction from that mission.
(1)load/save metadata defs
(2)load/save the data selection
(3)unify/align .dcp and.dcv files, and probably some other files I have not seen.
I kind of see a a final goal of DataCleaner to capture either an ETL process or a 'fixit-bot,' and my particular need seems to be a distraction from that mission.
A quick-and-dirty solution to the problem could be to just include a checkbox in the "Open CSV file" dialog that says "enable automatic narrowing and transformation of column types" and then pass "true" to "transformColumnTypes" parameter of DataContextFactory.createCsvDataContext(see javadoc here)...
I'm not sure if I understand the three ideas you posted, but I suggest you elaborate on it on the dev mailing list.
Log in by clicking the login link at the top of the screen
Go back to forum.
