[policy] Deprecate aggregator.module in D9 core and remove it in D10

We could improve aggregator.module and make it behave more like Feeds.

After having seen the aggregator module on the kill list over at #1255674: [meta] Make core maintainable i took a fist step at removing aggregator - an itch I have felt in a long time. The patch removes all the obvious files and fixes the test that broke. I have not touched system.api.php because I don't know how to make an example that good with another module. That still needs work.

With the patch applied, the only mentions of aggregator are in CHANGELOG.txt and in talking about aggregators in general.

What should happen to the module? I think that if someone wants to take over aggregator, improve it, and maintain it in contrib that would be great. If not then people will use the Feeds module.

We might want to put in an upgrade path from Aggregator to Feeds, or something similar in contrib, before removing this. There's a little site called the Drupal Planet that needs this.

Framework Initiative!

Yeah, I didn't really think of the Planet.. I guess it would be easiest to put aggregator in contrib and upgrade to that. But the nice solution would be to make an upgrade path to feeds. I'll take a quick look at how an upgrade to Feeds can be done.

Having any idea right now how an upgrade path to Feeds in D8 will look is maybe a bit unsure. If we want to get rid of aggregator in D8, I say move it to contrib - a little bit like what happened to blogAPI.

While a move from core to contrib wouldn't be bad, I find serious fault in thinking that Feeds is an upgrade path. For one Feeds isn't even stable, it's still in alpha for 7.x. The other problem is there's no development going on outside of a few bug fixes here and there. There's definitely no plan for an 8.x release yet. Nevermind the giant leap in complexity going from aggregator to Feeds.

This doesn't seem like a good idea to me. Being able to aggregate things from other websites is a very basic site building tool. I really don't want Drupal 8 to be released and 12 months later there's still no working aggregator solution for it...

As for Feeds, it's very good of course, but for now at least it seems like it would be extreme overkill for many sites that just need basic aggregation features.

OK, but what about pushing aggregator in contrib so user can choose simple vs advanced use case?

Being able to aggregate things from other websites is a very basic site building tool. I really don't want Drupal 8 to be released and 12 months later there's still no working aggregator solution for it...

IMO, those two statements are contradictory. If feed aggregation is a basic site building tool, then it won't take 12 months for said tool to be ready. Plus, what sites really use Aggregator, other than Planet? It seems like most sites are leaning toward Feeds, so we should remove aggregator from core and perhaps repurpose it as an easy frontend to Feeds in contrib (similar to simpleviews)

Can I please request that we stop these "Remove X from core" discussions and focus on getting D8 done?

Why is removing a module any less productive than spending a bunch of time on it for CMI updates, bug fixes, compatibility fixes, etc?

I am gonna mark this as duplicate of #1827164: Integrate Feeds into core

#15: The other issue depends on this one, and it isn't for 8.x.

Raising priority since #1930274: Convert aggregator processors and parsers to plugins needs an answer here.

#1930274: Convert aggregator processors and parsers to plugins Does not need any answer, job is already done, someone is needed to RTBC
#15266: Replace aggregator category system with taxonomy Job is almost done as well, waiting for taxonomy term replacement from entity reference patch
So if you think aggregator is useless and should be removed, stop by those issues and give a hand:)

If you think its not duplicate fine..but this is still filled as feature request, which automatically makes it subject to 9.x since we are in feature freeze

due to #1930274-44: Convert aggregator processors and parsers to plugins

Marking the Feeds in Core issue as a duplicate makes sense, but this issue is for completely removing Aggregator. I don't think that's a duplicate..

Just stumbled on this, and while I couldn't agree more with webchick, to not allow discussions on the future impede getting d8 done; i also agree with the comment in #10

This doesn't seem like a good idea to me. Being able to aggregate things from other websites is a very basic site building tool. ... As for Feeds, it's very good of course, but for now at least it seems like it would be extreme overkill for many sites that just need basic aggregation features.

After d8 has found its feet and the contributors that make it happen have recovered their breath.

I would certainly speak to the merits of aggregator even though there are a few features I find lacking, it does fill a special niche although I am not sure how the initiative to bring feeds into core will shape up.

Moving to the ideas queue.

Based on the usage data we have: #2867597: Top Drupal 7 and Drupal 8 core sub-modules the case for removal could certainly be made.

Adding the "idea template" to the issue summary. Short answers to the 4 questions will help us compare and weigh ideas for their relative impact. Thank you!

I've updated the issue summary, including details of how Aggregator relies on one of the dependencies that has prevented full PHP 8 support in Drupal 9.1.

Thanks @catch, that reads as Aggregator obstructing the whole of core in keeping with the times, as far as PHP is concerned.

Time to get hands-on and actually do this? (are there issues/docs that document how this process works?)

Agree with @yoroy. I think criteria for core module removal should include (a) maintainership (b) whether the given module has an ecosystem, where moving it to contrib may make it difficult to foster that ecosystem due to competing solutions showing up versus whether the module is a standalone feature (c) usage of the module as much as we can measure.

As far as I see, aggregator scores bad in all three aspects. So IMHO Drupal 10 would be a good time to remove. Much better solutions exist in contrib, so no wonder people are not using the core module too widely. It could still be maintained in contrib by people who care for it, so the functionality would not go away if there is interest in it.

I'd put my hand up to maintain this if it moved to contrib

@larowlan +1 as co-mantainer! I'm often using the module as example for scheduler

I think it needs to check with feeds roadmap (there's beweekly meetings)

Given that this ideas issue now has favorable feedback from two Product Managers, (@Gábor Hojtsy and @yoroy), plus willing contrib maintainers, we can probably go ahead with this.

Removing a module from core is supposed to have all of Product, Framework, and Release Management signoff. I think the fact that @larowlan offered to maintain it in contrib probably also implies his signoff as a Framework Manager? And from the Release Management side, I'm definitely in favor of doing this given the low module usage and the opportunity to remove a core dependency for D10. Pretty sure @catch is in favor as well given that he resurrected the issue in the first place.

So, I'm marking this policy decision RTBC. I'll do a quick committer poll just to confirm I haven't misrepresented anyone. After that, I think we can open a new core issue to follow through and deprecate the module.

IMO the next step after that will be to solve #3135100: [policy and meta] Provide a proper mechanism for deprecating modules and themes so that we can do the deprecation efficiently (rather than the sorta-frantic and inconsistent way we did module deprecations in the lead-up to 9.0.0). We can postpone our core issue for Aggregator on that soft-blocker.

Aright, only positive responses for the past week. I've filed #3188549: [Meta] Tasks to deprecate both Aggregator and our dependency on Laminas Feed for the next step and we can call this fixed. Thanks everyone!