Ian Ward

@kowh-ai looks like #136 resolves this issue. Maybe we should consider it for both the docker-compose files?

site_url shouldn't have path components, that's what [root_path](http://docs.ckan.org/en/latest/maintaining/configuration.html?highlight=configuration#ckan-root-path) is for. The url_for helper can be used to generate full paths correctly, I think.

It's being used because these are used as postgres column names in the datastore database, which can be difficult to access for users not used to entering non-ascii text via...

The difficulty I'm thinking of is for people without input methods set up to enter non-ascii characters reliably, and the ambiguity of some Unicode characters. It's not a blocker just...

@aimalkhan sounds good, but we'll need to check that the column name will be valid as a postgres column name (at a minimum no `"` characters but I'm sure that...

@jqnatividad for the alphanumeric test are you using the unicode categories `L*` and `N*`?

@duttonw I've sent you a pypi invite. @ThrawnCA do you have a pypi account?

When re-xloader-ing data if we use `datastore_records_delete` instead of `datastore_delete` (or `datastore_delete` with an empty filters dict) then the existing columns and data dictionary values won't be removed. Then `datastore_create`...

Note that the new IDataDictionaryForm allows users to customize their data dictionaries with private data that may not be appropriate for a public schema in the resource. So for now...

In fact xloader should add its own settings to columns with an IDataDictionaryForm so things like "strip whitespace" will be presented in the data dictionary form and validated by ckan.