Paragraphs and Default Content

+++ b/src/Controller/ParagraphsPreviewController.php
@@ -0,0 +1,39 @@
+		die($paragraph);

Tab in here need to be removed.

Looks promising

There was initial effort to base the entity reference dependency calculation on the UUID. We should try to revive that, as that would reliably work for any entity type, unlike the URL.

Either that and/or fix rest/serialization in core to provide a default pseudo-URL (like entity/entity_type/ID) for entities without a canonical URL. What it does right now is just bogus and only works because it uses old, deprecated API's.

This is the wrong approach IMHO and will have weird side effects. Generic API's will try to link to paragraphs then, build corresponding routes and so on.

We also found a second problem, and that's that saving apparently creates a new revision id and then the reference doesn't match anymore.

I'll ping @larowlan about this issue.

Can you provide some more detail about the saving/reference revision issue?
Do we need a ->setNewRevision(FALSE) or some-such on the paragraph?

I used to export/import paragraphs without any issues, so no more fixes needed

The problem mentioned by berdir in #8 still occurs. I think it needs fixing in Entity Reference Revisions, see #2780395: ERR fields are not deployable.

The patch in #4 could not be applied anymore due to a change in paragraphs.routing.yml.

Fixed file location and spacing errors.

This patch is not needed for default content and will not be committed like that.

Consider using #2848878: Embed Paragraph Content in Normalized Parent Entities (REST) instead, which allows to export a host entity including all its nested paragraphs.

As I reported there as well, the combination of default content and that patch doesn't work too well. I think the problem lies in my use of dcer to export the content, which then also exports the paragraph. Subsequently importing both the content with the embedded paragraph as well as the seperate paragraph then fails with an integrity constaint volation.

Sorry to re-open, but I think pointing to the ERR issue is a bit too quick. We've tried using that patch, but quickly run into problems when the target environment already has paragraphs and thus gets overlapping IDs. Not sure if ERR should not be exporting IDs, or whether default content should take action to not create overlapping IDs/ignore the IDs generated by the ERR normalization.

Adding to the above: maybe it is an option to keep this open at least as long as #2848878: Embed Paragraph Content in Normalized Parent Entities (REST) hasn't been finalised to keep track of the effect on default content and guard the progress on getting this to properly work in DC? Maybe with status postponed?

That is how default_content works, there is an issue there to change that but until then, I think the patch should stay consistent with the current behavior of the module (you can also get into conflict on node ids).

And if we change it, we have to change it there, there's no point in keeping this open.