datasetType (software, workflow, review, etc.) · dev

But now that I'm converting allowedDatasetTypes from a string to a ManyToMany list, things are getting more complicated. I need to be able to unset the list.

Philip Durbin 🚀 (Feb 12 2026 at 15:23):

That's what we did at the dataset level when we introduced dataset types. All existing datasets were set to the "dataset" dataset type (using Flyway).

Oliver Bertuch (Feb 12 2026 at 15:25):

Here's an idea: what if for now we make the list non-optional and a migration adds the minimum default to all. At a future point in time, deleting the list and replacing it with null or empty would mean "inherit"

Oliver Bertuch (Feb 12 2026 at 15:25):

So we can add any other calls that are necessary to do that in a new iteration

Philip Durbin 🚀 (Feb 12 2026 at 15:26):

Actually, that gives me an idea. I was trying to unset the list but maybe I can just explicitly set the list to a dataset type that I'm not trying to delete.

Philip Durbin 🚀 (Feb 12 2026 at 15:28):

At the moment, I'm just trying to get DatasetTypesIT to pass along with the change from a String to a List for allowedDatasetTypes.

Philip Durbin 🚀 (Feb 12 2026 at 15:32):

Yes! It works! :tada:

Oliver Bertuch (Feb 12 2026 at 15:33):

Yeah my suggestion about not allowing "null" or empty list for now was going in this direction of "set the list to a type that still makes sense"

Oliver Bertuch (Feb 12 2026 at 15:34):

What happens if you delete a type itself that has been used? It shouldn't be allowed to be dropped, right?

Philip Durbin 🚀 (Feb 12 2026 at 15:35):

"message": "java.lang.IllegalStateException: Dataset type with id 6 is referenced and cannot be deleted."

Philip Durbin 🚀 (Feb 12 2026 at 15:35):

but now I have the workaround above, which I'm documenting

Oliver Bertuch (Feb 12 2026 at 15:35):

Referenced by whom? The collection or a dataset?

Oliver Bertuch (Feb 12 2026 at 15:36):

IMHO that is an important distinction.

Philip Durbin 🚀 (Feb 12 2026 at 15:37):

Well, this worked before so I'm assuming it's the new List relationship to the collection.

Philip Durbin 🚀 (Feb 12 2026 at 15:42):

Woo-hoo! Tests passing! I'll clean this up, commit, and push.

Philip Durbin 🚀 (Feb 12 2026 at 16:39):

@Oliver Bertuch is this incrementally better? :sweat_smile: https://github.com/IQSS/dataverse/pull/11753/changes/d2f12dfd9eec3d782e68493b6f9293fecaf60b9e - switch allowedDatasetTypes in entity from String to a ManyToMany List

Oliver Bertuch (Feb 12 2026 at 16:41):

I need to look into our new storage option that (finally) came through today, so won't be able to look now.

Philip Durbin 🚀 (Feb 12 2026 at 16:43):

No worries, I'll look through other stuff to fix that we talked about. Thanks again!

Philip Durbin 🚀 (Feb 25 2026 at 16:38):

@Oliver Bertuch I just noticed you still have requested changes on https://github.com/IQSS/dataverse/pull/11753 . Is that right? Or should I click "dismiss review"?

Also, @Gustavo Durand picked up that PR to review it.

Oliver Bertuch (Feb 25 2026 at 16:39):

I have not looked at the remarks and the changes you made again. Maybe Gustavo can decide what he thinks about them.

Philip Durbin 🚀 (Feb 25 2026 at 16:44):

We might have a zoom call if you'd like to join. Not sure when.

Oliver Bertuch (Feb 25 2026 at 16:44):

Sure. Today would be a good fit for that.

Philip Durbin 🚀 (Feb 26 2026 at 14:01):

I made those changes for you :heart:

Oliver Bertuch (Feb 26 2026 at 14:03):

Let me take another look. Posters are printed. :smile:

Oliver Bertuch (Feb 26 2026 at 16:08):

Done.

Philip Durbin 🚀 (Feb 26 2026 at 16:23):

Thanks!!

Oliver Bertuch (Feb 26 2026 at 16:24):

I did my best to reply to open questions and found a few other issues.

Oliver Bertuch (Feb 26 2026 at 16:24):

The most important one is certainly the missing equals() and hashCode() on DatasetType.

Philip Durbin 🚀 (Mar 02 2026 at 20:54):

@Oliver Bertuch is just id and name enough? This is what VS Code is suggesting:

Screenshot 2026-03-02 at 3.42.27 PM.png

If I check just id and name, it shows generates this:

    @Override
    public int hashCode() {
        final int prime = 31;
        int result = 1;
        result = prime * result + ((id == null) ? 0 : id.hashCode());
        result = prime * result + ((name == null) ? 0 : name.hashCode());
        return result;
    }

    @Override
    public boolean equals(Object obj) {
        if (this == obj)
            return true;
        if (obj == null)
            return false;
        if (getClass() != obj.getClass())
            return false;
        DatasetType other = (DatasetType) obj;
        if (id == null) {
            if (other.id != null)
                return false;
        } else if (!id.equals(other.id))
            return false;
        if (name == null) {
            if (other.name != null)
                return false;
        } else if (!name.equals(other.name))
            return false;
        return true;
    }

Philip Durbin 🚀 (Mar 02 2026 at 20:57):

Anyway, pushed: https://github.com/IQSS/dataverse/pull/11753#discussion_r2874633026

Philip Durbin 🚀 (Mar 02 2026 at 21:50):

And this: https://github.com/IQSS/dataverse/pull/11753/changes#r2874761915

Oliver Bertuch (Mar 02 2026 at 21:52):

Sorry I was out with @Jan Range in Stuttgart. We had a fab diner at a local Brauhaus!

Oliver Bertuch (Mar 02 2026 at 21:52):

Do you still want input?

Philip Durbin 🚀 (Mar 02 2026 at 21:52):

For the other stuff, @Gustavo Durand @Juan Pablo Tosca Villanueva and I decided we'd punt. I'm going to create a tech debt issue for dataset types. Maybe you can help edit the description, @Oliver Bertuch, and we'll decide on the scope and size of the next push, whenever it is.

And I'm still open to adding small things to the PR we're working on now.

Oliver Bertuch (Mar 02 2026 at 21:53):

Sounds good to me! I'm just the messenger pointing at things, you decide the course, cap'n :man_rowing_boat:

Philip Durbin 🚀 (Mar 02 2026 at 21:54):

Ha. I'll let you know when I've created it.

Philip Durbin 🚀 (Mar 02 2026 at 21:54):

I literally wrote "create issue about corners that were cut" :sweat_smile:

Philip Durbin 🚀 (Mar 02 2026 at 21:55):

But on the plus side, hopefully we'll get #11753 merged in time for 6.10.

Philip Durbin 🚀 (Mar 03 2026 at 19:23):

@Oliver Bertuch I'm clicking "re-request review" on https://github.com/IQSS/dataverse/pull/11753

Screenshot 2026-03-03 at 2.22.59 PM.png

I don't want this to be a blocker for merging. (I know that worst case, we can always dismiss your review.)

Philip Durbin 🚀 (Mar 04 2026 at 12:42):

@Oliver Bertuch hi, just checking in. Merging is blocked:

Screenshot 2026-03-04 at 7.41.25 AM.png

Oliver Bertuch (Mar 04 2026 at 12:50):

Just today I had a request by a researcher asking for Instruments support. Any objections to add those? https://support.datacite.org/docs/can-i-register-a-doi-for-an-instrument

Oliver Bertuch (Mar 04 2026 at 12:51):

(Not as a part of this PR)

Philip Durbin 🚀 (Mar 04 2026 at 12:52):

Any types can be added freely.

Philip Durbin 🚀 (Mar 04 2026 at 12:53):

Just cook up a JSON file.

Oliver Bertuch (Mar 04 2026 at 12:53):

But we'd need to extend the enum for resourceTypeGeneral, don't we?

Philip Durbin 🚀 (Mar 04 2026 at 12:53):

Well, sure.

Oliver Bertuch (Mar 04 2026 at 12:53):

Like you did for reviews

Philip Durbin 🚀 (Mar 04 2026 at 12:54):

If you want it to be sent to DataCite or show up in the CSL citation export.

Oliver Bertuch (Mar 04 2026 at 12:54):

Yeah, I'd like that. DataCite has native support for Instruments these days and it would be great to keep the data and instruments in the same place.

Philip Durbin 🚀 (Mar 04 2026 at 12:56):

Sure. Future PR, though, please! :sweat_smile: Code freeze is tomorrow.

Oliver Bertuch (Mar 04 2026 at 12:56):

I think I said that above :wink:

Philip Durbin 🚀 (Mar 04 2026 at 12:56):

So we're trying to get this one tested and merged.

Philip Durbin 🚀 (Mar 04 2026 at 12:58):

Are you still thinking about the "merging is blocked" question above? :heart:

Philip Durbin 🚀 (Mar 04 2026 at 12:58):

You sort of avoided talking about it. :smile:

Oliver Bertuch (Mar 04 2026 at 12:58):

Sorry at a conference.

Oliver Bertuch (Mar 04 2026 at 12:58):

I am just doing the review in parallel to a session

Philip Durbin 🚀 (Mar 04 2026 at 12:58):

Worst case we'll dismiss your review so we can merge.

Oliver Bertuch (Mar 04 2026 at 12:58):

(As you may have noticed)

Oliver Bertuch (Mar 04 2026 at 13:02):

I just revisited my main concerns and find them all covered now. Approved. :check: Thanks for working on this and bearing with my lengthy remarks.

Philip Durbin 🚀 (Mar 04 2026 at 13:40):

Sure! To be honest, I need to do a little more testing, so I'll be back on that branch when I get to work.

Philip Durbin 🚀 (Mar 04 2026 at 13:40):

So if there's anything small you want, lemme know.

Juan (Mar 04 2026 at 13:53):

Philip Durbin 🚀 said:

Sure. Future PR, though, please! :sweat_smile: Code freeze is tomorrow.

In Spain -and I’ve heard also in Latin America, and I think in France as well-, we are receiving petitions to upload Data Management Plans to our Dataverse instances.

Could Data Management Plan be considered a new dataSetType? Should I open a new PR?

Philip Durbin 🚀 (Mar 04 2026 at 14:30):

@Juan interesting! I'm think we should have a different topic about DMPs. Do you want to create one? Maybe in #community? I'm sure @Viviane Veiga and many others will be interested. By the way, in #12000 we're adding this:

As shown in a video <https://www.researchspace.com/blog/dmptool-rspace-eln-dataverse-integration>, a Data Management Plan (DPM) can be added into RSpace and the research records and associated data can then be sent to Dataverse. Dataverse generates a Persistent Identifier (PID, often a DOI) for the dataset, and RSpace automatically puts the PID link under "Research Outputs" in the DPM.

(Thanks to @Tilo Mathes and @Vaida Plankytė 🎨 and the rest of RSpace for that integration!)

Juan (Mar 04 2026 at 14:42):

Thanks @Philip Durbin 🚀, I will watch the video and open the new topic tomorrow. I have meetings for the rest of my workday.

Oliver Bertuch (Mar 04 2026 at 15:04):

@Juan @Philip Durbin 🚀 DataCite has support for a "OutputManagementPlan" type these days.

Philip Durbin 🚀 (Mar 04 2026 at 15:04):

I saw that. Interesting.

Oliver Bertuch (Mar 04 2026 at 15:05):

@Philip Durbin 🚀 should we discuss in a new GH issue if the resource type general should be a configurable type and not hardcoded?

Philip Durbin 🚀 (Mar 04 2026 at 15:06):

Or maybe a design doc? I started this one called "Resource Types" back in October: https://docs.google.com/document/d/1xUbOj0vQ629_YF0nMM3OUnfjY4QBw3WVsD_MnrMMBYI/edit?usp=sharing

Oliver Bertuch (Mar 04 2026 at 15:06):

With output management plans comes to mind that there are DMPs and SMPs and maybe more in the future.

Oliver Bertuch (Mar 04 2026 at 15:06):

Sure.

Philip Durbin 🚀 (Mar 04 2026 at 15:06):

(I don't find GitHub issues to be great for long discussions.)

Oliver Bertuch (Mar 04 2026 at 15:06):

It would be good to collect use cases and think some more about where we wanna go with this

Oliver Bertuch (Mar 04 2026 at 15:07):

Should we keep the existing design doc and add another chapter/part to it?

Philip Durbin 🚀 (Mar 04 2026 at 15:08):

Well, remember we have an even earlier design doc: Proposal: Supporting Multiple Dataset Types in Dataverse - https://docs.google.com/document/d/16RvGXmaPQK9DGsEEbrrFEu8mjUrN93YY_yh7ZONyMDI/edit?usp=sharing

Oliver Bertuch (Mar 04 2026 at 15:09):

Oh so we already have multiple. We can always link between them for tracking the history

Philip Durbin 🚀 (Mar 04 2026 at 15:09):

I'm fine with whatever. New doc. New Zulip topic. A meeting. Anything but trying to have a discussion in a GitHub issue. :smile:

Notification Bot (Mar 04 2026 at 15:17):

5 messages were moved from this topic to #dev > Architecture Decision Records (ADRs) by Oliver Bertuch.

Oliver Bertuch (Mar 05 2026 at 09:06):

So sorry @Philip Durbin 🚀 but I did find a small thing. :see_no_evil: Stumbled over this just this morning while looking for other things in the DataCite rTG vocabulary. https://github.com/IQSS/dataverse/pull/11753/changes#r2888545716

Philip Durbin 🚀 (Mar 05 2026 at 13:13):

@Oliver Bertuch thanks, I just replied

Philip Durbin 🚀 (Mar 05 2026 at 14:05):

Want me to add a comment in the code?

Oliver Bertuch (Mar 05 2026 at 15:23):

That may be good, as others may have the same question. Should this go into the docs as well?

Philip Durbin 🚀 (Mar 05 2026 at 15:32):

Maybe. Where?

Oliver Bertuch (Mar 05 2026 at 16:00):

On the train ride home now. Let me take a look.

Oliver Bertuch (Mar 05 2026 at 16:19):

I left a few (three) suggestions on the user doc.

Philip Durbin 🚀 (Mar 05 2026 at 17:07):

@Oliver Bertuch I don't suppose you have time for a call

Philip Durbin 🚀 (Mar 05 2026 at 17:07):

The name is "dataset" and the displayName is "Dataset". Maybe I can edit your suggestion to make this clear.

Oliver Bertuch (Mar 05 2026 at 17:08):

In a public no-phone environment at the moment... Sorry. Will be back to desk later if that helps.

Oliver Bertuch (Mar 05 2026 at 17:08):

I was suggesting to use "Dataset" as this is the user documentation and users will see "Dataset" in the UI

Oliver Bertuch (Mar 05 2026 at 17:08):

Feel free to ignore any suggestions that don't resonate with you

Oliver Bertuch (Mar 05 2026 at 17:08):

I don't feel strongly about any of this.

Philip Durbin 🚀 (Mar 05 2026 at 17:09):

Actually, since that suggestion isn't in the API Guide, I'll take it.

Philip Durbin 🚀 (Mar 05 2026 at 17:14):

Actually, I took it but I'm going to change it back to lower case "dataset" because that's what is shown in the search facet (as of this PR).

Philip Durbin 🚀 (Mar 05 2026 at 17:26):

Ok, I think I'm done making changes: https://github.com/IQSS/dataverse/pull/11753/commits

Philip Durbin 🚀 (Mar 05 2026 at 17:27):

Juan said:

Thanks Philip Durbin 🚀, I will watch the video and open the new topic tomorrow. I have meetings for the rest of my workday.

@Juan thanks for kicking off #community > A new supported datasetType for Data Management Plans? !

Philip Durbin 🚀 (Mar 05 2026 at 17:29):

@Juan since you're obviously using dataset types (or plan to), I wanted to give you (and others) a heads that in #11753 the following backward-incompatible change is coming:

In previous releases of Dataverse, as soon as additional dataset types were added (such as "software", "workflow", etc.), they could be used by all users when creating datasets (via API only). As of this release, on a per-collection basis, superusers must allow these dataset types to be used. See #12115 and #11753.

Juan (Mar 05 2026 at 20:22):

Thanks @Philip Durbin 🚀 , We will migrate to v6.10 before starting to use dataset types.

Philip Durbin 🚀 (Mar 05 2026 at 20:24):

Ok, phew :relieved:

Philip Durbin 🚀 (Mar 05 2026 at 20:24):

Longer term, maybe dataset types should turn into resource types. I don't know.

It would be an architectural change. I gave you comment access on the Resource Types doc I mentioned above. But that doc need a lot more work.

Juan (Mar 05 2026 at 20:27):

Perfect. Thanks

Philip Durbin 🚀 (Mar 05 2026 at 20:28):

Have you played with Zenodo? They allow you to create a bunch of different resource types.

Juan (Mar 05 2026 at 20:30):

A little bit, but we like much more Dataverse :smile: . But we will need to add at the moment DMPs, and software. Perhaps another resource types in the future.

Philip Durbin 🚀 (Mar 05 2026 at 20:43):

Here's the Zenodo UI:

zenodo.png

Philip Durbin 🚀 (Mar 05 2026 at 20:50):

The OSF UI:

osf.png

Juan (Mar 05 2026 at 20:50):

It seems that they do not have a specific resource type for DMPs. Even the DMPs imported from Argos have "Other" resource type: https://zenodo.org/records/3980771 . Argos and Zenodo are OpenAIRE tools.

Philip Durbin 🚀 (Mar 05 2026 at 20:53):

They both seem to have the option for Output Management Plan (which, if you squint, might look like a DMP :smile: ):

omp.png

Philip Durbin 🚀 (Mar 05 2026 at 20:54):

Any objections if I move all these messages into the main topic (for dataset types)?

Update: moved! Sorry if it's messy! I moved the messages about potential DMP and instrument dataset types here.

Juan (Mar 05 2026 at 20:55):

Yes, you are absolutely right. I didn't find it under publication.

Philip Durbin 🚀 (Mar 06 2026 at 15:31):

@Oliver Bertuch @Gustavo Durand I just posted more about nullable true vs false for displayName at https://github.com/IQSS/dataverse/pull/11753/changes#r2788164627

Philip Durbin 🚀 (Mar 06 2026 at 15:32):

I might take another run at making displayName nullable=false. Wish me luck!

Philip Durbin 🚀 (Mar 06 2026 at 16:07):

Wait, wait, it's already nullable=false:

@Column(nullable = false, columnDefinition = "VARCHAR(255) DEFAULT ''")
private String displayName;

@Oliver Bertuch is there anything to fix here? Or in the following?

-- Add displayname column to datasettype table
ALTER TABLE datasettype ADD COLUMN IF NOT EXISTS displayname VARCHAR(255) NOT NULL DEFAULT '';
-- Populate displayname with name but capitalize it (name=dataset becomes displayname=Dataset)
UPDATE datasettype SET displayname = CONCAT(UPPER(SUBSTRING(name, 1, 1)), SUBSTRING(name, 2));

Last updated: Apr 03 2026 at 06:08 UTC