The documentation, Updating your metadata - Crossref , mentions that to update the Crossmark metadata (any of its fields, I guess?) we need to make two-step deposits. But I’m curious as to why this is so. Is it some technical thing related to the inner workings of Crossref? I’d appreciate if someone knowledgeable could answer a few questions in this regard.
Is two-step deposit a hard restriction that must always be followed? What happens if we forget about two-step deposits and upload a full-record XML with the Crossmark metadata that is different from that in the initial deposit? Will this result in an error? Or will the old Crossmark just not be overwritten by the new data?
In other words, can the Crossmark-only-XML deposits be used to update the Crossmark, or only to add the Crossmark data to the records that don’t have any previous Crossmark data?
We have a lame XML generator that produces Crossmark-less Crossref XMLs. To participate in Crossmark, we can make a separate lame XML generator function that would make Crossmark-only-XMLs. Now, we’ll have to upload Crossref XML and then Crossmark-only XML to register all the data. But what happens when we need to update the old record? Will uploading the Crossmark-less Crossref XML delete the existing Crossmark data associated with the record? Will we also have to redeposit the Crossmark-only XML for the updated record?
Hello, and thanks for bringing that to our attention.
That documentation is phrased in a confusing way. The two-step process (i.e. delete all the existing Crossmark metadata by using an empty, self-closing <crossmark/> tag before adding new metadata) is not necessary.
If you submit a metadata update (either as a full metadata deposit or resource deposit) for Crossmark metatdata, the <crossmark> section in the update will entirely overwrite the whole prior <crossmark> section from previous metadata deposits.
I’ll work on getting the documentation clarified and updated.
Thank you for the answer! That’s actually great news! Seems we can treat the Crossmark metadata just like any “regular” metadata, such as abstracts or URLs, without extra complexity associated with depositing and updating Crossmark.
Yes, that’s generally true. It is a little bit different than abstracts, but similar to things like URLs, references, licenses, funding data, and relationships.
The difference is that abstracts are part of the main body of bibliographic metadata (along with titles, authors, publication dates, etc., etc.). And as such they can only be added or updated in a full metadata deposit.
Those other ancillary metadata elements, including Crossmark metadata, can be updated on their own in resource-only deposits, or as part of a full metadata deposit.
What this means in practice is twofold.
If you have a record with Crossmark metadata already in it, and then you submit a full metadata deposit which does not have Crossmark, that will not serve to remove the preexisting Crossmark metadata. It keeps on existing in its own little bubble, and has to be proactively deleted (by means of a self-closing empty <crossmark/> tag) if you want it removed.
But, if you do submit updated Crossmark metadata (either in a full deposit or in a resource deposit) the whole <crossmark> section will be overwritten with whatever you just submitted. You can’t incrementally add to it. Just like you can’t submit half a reference list and then add in just the other half with a new resource deposit of references. The second half would replace the first half in that case, not be appended to it.