Anonymous failed to validate ... Name (name): You must enter a username

It looks like that since Feeds was trying to update the anonymous user, the source column you specified on the mapping did not match with the CSV file.

• If you choose to use no headings for your CSV file, your source columns you should be named 0, 1, 2, etc. on the mapping page.
• When you do use headings, the headings of your CSV file have to match exactly with sources specified on the mapping page. Note that in the D8 version of Feeds, the sources names are also case sensitive. So the CSV headings don't match with the mapping sources if you specify them with a capital in one place and in lowercase in the other.
• In order to successfully import new users, username and mail address are required. So a CSV file with a single column for importing new users won't work. For updating existing users you don't need to have both in your source file.

Suggestions for making above more clear in the UI are welcome. :)

Ah, I see you tried to set a path as source on the mapping form.

Providing the source file happens at a later stage, when creating a feed.

On the mapping form, when using the CSV parser, you need to enter the column names used in your CSV file, not a path to the CSV file itself.

Example:

What do you mean "since Feeds was trying to update the anonymous user"?

The error message said: "Anonymous failed to validate". This gave me the hint that Feeds was trying to update user 0, the anonymous user in Drupal. That user has no username. Apart from that Feeds shouldn't try to update that special user, that additionally gave me the hint that the username Feeds received from the source must be empty, thus resulting into Feeds loading the only user that has no username: the anonymous user. And since Feeds received an empty source, that gave me the hint that something on the mapping page was not configured as expected.

How do I "named 0, 1, 2, etc. on the mapping page"? I don't see that option.

Add a new CSV source and type in "0".

Again, suggestions for improving the user interface are welcome!

@sajosh
Thanks for your suggestions! The case sensitivity could differ per parser. And for not every parser you need to enter sources. For some parsers you only need to select from predefined sources, as is the case for the RSS parser. Sources may also mean slightly different things per parser. For example, for the XML parser (from Feeds extensible parsers) they are not fields or headings, but XPaths. And not all CSV sources have to be column names. You can also select from predefined sources. For example: you could add a field to the feed type and map that to a target field.

So as a start, what if we only add help text for the CSV parser now? See images below how that would look like and the patch for the implementation.

No headers

With headers

I would love your feedback! :)

@sajosh
You are making some nice suggestions. Unfortunately, we cannot change the CSV no header start with 1, as that could possibly break existing configurations that now use 0 as mapping source. So I've noted that in #2984386: Build working UI for csv import for the Feeds Migrate project, which aims to become the next generation Feeds module.

Target on the left

We could put the Target column first, but don't people usually think from left to right? So left where things are coming from and right where things are going to?

In the blank box that appears after selecting New CSV source, can you put "type case-sensitive header name or 0,1,2 if no header"

Nice suggestion, though I have to think about how to properly implement that in a way that it doesn't affect other parsers.

I yet have to look to your suggestions on the instructions. I didn't wrote the instructions on https://www.drupal.org/docs/8/modules/feeds/creating-and-editing-import-....

Thanks!

What are "Feed: " list?

"Feed: " mapping sources are fields from the feed entity. You could use this for example in this way:

1. Add a text field to the feed type: /admin/structure/feeds/manage/my_feed_type/fields
2. The added text field appears a source. Map this to a target.
3. Create (or edit) a feed. Beside providing a source here, there is an additional text field in which you can fill in a value.
4. Fill in a value for the text field. All imported content will get that value in the mapped target field.

This is useful in for example the following scenario:

• You need to import multiple feeds that use the same structure. For example you have a Drupal site with store and each manufacturer need to provide their own CSV file in a format that you dictate. So each one needs to deliver the CSV file with the same set of columns.
• The manufacturer creates a feed (or has access to a single feed) and uploads their CSV file. In the text field they specify their name.
• The text field is mapped to a field called "field_manufacturer" on the product. Now each imported product will get the manufacturer's name on the product.

Now that I think about, in the D7 version of Feeds there was a legend, that explained each mapping source (and each mapping target). That's one of the things still missing in the D8 version:

@sajosh
I've added a patch for legend in #3074085: Display descriptions of sources and targets. This displays description for each source and each target.

Not sure yet how to get a description for a group of sources. Descriptions currently can only exist per source.

This builds further upon the patch from #6 and adds a description to the "New CSV source" field.

With headers

Without headers

@sajosh
Thanks for testing! I tested with caches disabled, so I wasn't aware of the fact that help text got cached. So the help text should not be dependent on the settings then (or I would need to figure out how to clear the right caches when the config changes). This means that the help text also remains the same when changing the parser plugin.

For now, I've removed the help text per plugin thing completely. I replaced it with a link to the docs as you suggested.
The description for the "New CSV source..." option does not have this problem so this remains dependent on the configuration. I did remove the quotes from "0, 1, 2" as you suggested. I think you are right someone could think they need to enter the quotes.

You've been of great help! :)

The mapping form:

I slightly changed the description for entering a new CSV source when using the "No headers" option. It is now:

Enter which column number of the CSV file to use: 0, 1, 2, etc.

The reason is that now more than a month later I read "Enter the number of the column" as if I need to enter how many columns the CSV file has. So I hope the new description is more clear, despite the slight ambiguity of the word "use".

Committed #14 with that change.

Thanks @sajosh for all your suggestions to improve the user experience.