Should Drupal 10 support migrating from Drupal 7?

Added Drupal 7 vendor extended support link/time which could IMHO be the deciding factor here.

I'm very heavily leaning towards:

1. Continue to support Drupal 7 migrations to Drupal 10 in core.

1. As long as we're within the D7 LTS it means we should stick to our promise to help pass through the migration roadblock, especially considering it should be the last one.
2. Since Drupal 8 was released, Drupal's usage kept shrinking, possibly because of the cost to migrate and/or learning curve/effort. A few months ago, Shopify even passed above Drupal...in the CMS usage trends (how the irony).
3. Speaking of which, if we don't make the upgrade experience as streamlined as possible, then we're at risk of seeing SMBs and Enterprise customers leave Drupal. This means less contributions and less investment flowing into the project, which isn't good for anybody. When it comes to evaluating the migration effort, challenging Drupal VS different technologies, re-train developers and allocate significant budget, we should make sure to have people's back as much as we can and have Drupal remain the best option.

@anavarre: well, removing migrations from Drupal 7 in Drupal 10 would mean that we keep maintaining the Drupal 7 to Drupal 9 upgrade path until the last Drupal 9 release (which should have the same data structure as the first Drupal 10 release). So you *could* use the contrib module to migrate from Drupal 7 to Drupal 10.0, as it would be entirely compatible with it. Even if it used deprecated code, that would not be the case anymore in the last Drupal 9 release per definition of how we mark deprecated code now. So removing it from Drupal 10 mean that you would still have an up to date migration path but Drupal 10 would not need to maintain it for years onwards.

Also we made the Drupal 7 LTS a commercial program because we did not have community resources anymore to maintain it. The same question applies to the Drupal 7 migration path really. With it being in Drupal 9 (even the Drupal 6 path being in there, see #3010983: Deprecate Drupal 6 and Drupal 7 migrations and move to contrib) we are already keeping it working up until the release of Drupal 10 at least already as it is now, so up until at least end of 2023 (end of support for Drupal 9).

you *could* use the contrib module to migrate from Drupal 7 to Drupal 10.0, as it would be entirely compatible with it.

In the event the Drupal 7 migration path would indeed be stable by the time D9 comes out, IDK how long it'll take before people actually test things on real sites and report back with meaningful data (and corresponding bug reports)? I anticipate we'll know exactly by mid-2021 at best, and possibly later via the LTS. Would it be an option to make our decision based on D7 usage then or will it be too late?

Also, I don't think it's an overstatement to assume if we put this in contrib, in the event criticals bugs are uncovered, we might not be prompt at taking care of them since it won't be a focus for Drupal core anymore (you did mention community resources, which will remain an issue, sadly).

Is it an option that instead of supporting migrating from Drupal 7 for the entire lifespan of Drupal 10 we'd agree to remove migrations in a minor release once D7 LTS is behind us? Or would that break backwards compatibility somehow and it would force us to only remove them in Drupal 11?

In the event the Drupal 7 migration path would indeed be stable by the time D9 comes out

Well, its a release requirement for Drupal 9 for that to be the case for core stuff. We cannot hold core for lack of contrib migrations which is probably what is going to hit actual sites.

Would it be an option to make our decision based on D7 usage then or will it be too late?

We did undeprecate things in Drupal 8 when it turned out it was not feasible to do what we wanted. So deprecating the module in Drupal 9 would not 100% mean we actually get to remove it if there was a good reason to keep it.

Also, I don't think it's an overstatement to assume if we put this in contrib, in the event criticals bugs are uncovered, we might not be prompt at taking care of them since it won't be a focus for Drupal core anymore (you did mention community resources, which will remain an issue, sadly).

Well, Drupal 7 will also be in vendor's hands at this point, not in the "community"'s hands. The fundamental question here is should the migration path also become "vendor extended"?

Is it an option that instead of supporting migrating from Drupal 7 for the entire lifespan of Drupal 10 we'd agree to remove migrations in a minor release once D7 LTS is behind us?

Currently the Drupal 7 vendor extended support program says "participation for at least for 3 years in the program", not that that would be the absolute limit. So there is a chance the Drupal 7 vendor extended support program runs longer. (The Drupal 6 LTS program is still running as of now for example). So cannot know for sure now if Drupal 7 vendor extended support will surely end at end of 2024 or not. That said, if it would end, it would be a backwards compatibility change to remove migrate from core in the middle of a major release cycle, so it would need to be kept until Drupal 11 yes.

Thanks for clarifying! 👌

The fundamental question here is should the migration path also become "vendor extended"

Ha! That's a very good point.

I do think we need to have a D7-D10 migration path, regardless of whether the migrations exist in core or contrib.

How much work was it to port the (non-multilingual) sources and destinations to D9? Were there any challenging issues?

Now that Drupal extended D7 EOL to Nov 2022, i think a 7 to 10 migration path will be needed.

https://www.drupal.org/psa-2020-06-24

Correct Drupal 7 EOL date.

How much complexity does the exsitence of the Drupal 7 migration path mean in core? Are there potentials of significant improvements if moved to contrib or lowering maintenance overhead from core or is the migration path more like a "what's done is mostly ok/stable"?

Unrelated:

end of 2024 (well after Drupal 9 EOL and a year and half into Drupal 10).

Correcting this to two and a half years. Middle of 2022 + 1.5years is just end of 2023, not end of 2024.

This was discussed at the Migrate Meeting 2021-07-15 attended by benjifisher, mikelutz, heddn, marvil07 and srjosh. The short answer is yes, Drupal 10 should support migration from Drupal 7. It was an easy decision because Drupal 7 EOL is after Drupal 10 expected release date. Plus we all we want to provide an upgrade path that is as simple as possible.

Thanks everyone for the discussion.

At some point, we will want to remove support for migrating from Drupal 6 and Drupal 7. We could open another planning issue for that. Will it be in Drupal 11?

For the record, @dinarcon and @quietone also attended the meeting mentioned in #19.

Drupal 11 seems like a good date to drop support. Thanks for keeping it around for Drupal 10, though. I think there will indeed be folks who wait until then to do their upgrades.

@catch, @webchick:

Thanks for the comments. During the meeting last week,

1. We discussed the pros and cons of removing D6 and D7 support in one or two steps. We did not come to a firm conclusion.
2. We talked about setting up some sort of meeting with interested core maintainers--I guess that means at least the two of you!--to discuss these questions.
3. Also, if we remove support for D6 and D7 (or move that support to contrib), should we consider adding something to core?

When deciding whether to add a migration feature to core or to a contrib module (often Migrate Plus) we have been using the criterion: is it needed by the core migrations? If most of the core migrations go away, what will the new criterion be?

When deciding whether to add a migration feature to core or to a contrib module (often Migrate Plus) we have been using the criterion: is it needed by the core migrations? If most of the core migrations go away, what will the new criterion be?

On this one, I feel that migrating from X to Drupal is a perennial problem that will persist even until Drupal 2000. :) So it makes sense to me that the criterion would shift to something like "does this make it easier / less buggy / etc. to migrate from X to Drupal"?

Happy to join a meeting to discuss the others!