Repository navigation
Description
This issue tracks the work to formally decouple the material and cupertino libraries from the core Flutter framework and establish them as standalone, first-party packages on pub.dev.
➡️ 📄 Read more about our plan. ⬅️
This strategic transition will make Flutter's core lighter and more flexible, accelerate design library updates, and give developers more control over their app's dependencies.
The work will be managed through a deliberate, three-phased approach to ensure a stable and predictable migration for the entire ecosystem.
The sub-issues listed below track the specific tasks for each phase of this project.
📣 To provide feedback:
- Generally? Here.
- On critical refactoring of code out of material/cupertino before decoupling:
- On net-new additions of core components:
- On the new packages:
- On future deprecation and removal:
OP
Use case
- Long update cycle
- To get the new version - you need to update the version of Flutter
- Random imports
- Components for other platforms are separate from Flutter (fluent, macos_ui, yaru, ...)
Proposal
Move material and cupertino packages from Flutter
It can be done along with this issue (#91605)
Update 04/18/25
Link to proposal by @MaherSafadii with examples.
Metadata
Metadata
Assignees
Labels
Type
Projects
- StatusShow more project fieldsNo status
Activity
maheshj01 commentedon Apr 7, 2022
Hi @nikitadol, Thanks for filing the issue. If I understand it right moving material and Cupertino outside of flutter means, the flutter SDK wouldn't have any of the default material and Cupertino widgets, unless they are explicitly imported from pub.
I believe doing so would lead to inconsistency and duplication of code across both packages.
the packages that you linked are third-party packages so they are not part of the framework.
Labeling this issue for further thoughts from the team.
cc: @rami-a
Edit: The inconsistency and duplication I was referring to is the code which is being reused across both material and cupertino
nikitadol commentedon Apr 7, 2022
The
materialpackage depends oncupertino- I do not quite understand what the problem will bea-eniou-pvotal commentedon Apr 20, 2022
I think about this case and I found it is a good idea but in the long-term, it will lead to a fragmented community
Like a company uses AmazonButton and a different company using AppleButton so changing a company to the next company that uses different UI names and parameters will be a bad experience
LasseRosenow commentedon Apr 21, 2022
Except that scenario is very unlikely to happen, why would a company change its flutter app to use a totally different design system? Also right now this is not possible either, as cupertino and material have different widgets as well.
There are community efforts to bring together multiple design system in a platform agnostic way i think they are called
platform-widgetsor something like that.I think not. I think the best Idea would be to have basic stuff inside flutter, like
RawScrollbaretc. But everything that is different should be in a different package. Similar to how it is right now, but just that the package does not live inside flutter, but outside, which would have multiple advantages:LasseRosenow commentedon Apr 21, 2022
I think in general flutter has a very weird approach with including a gui library within the framework. It is very unusual.
Competing modern frameworks don't do it either (I know this is a weak argument, but I wanted to mention it :D). Like react or react native or lit or vue.js etc. They all are just frameworks that give you a way to implement what ever design you want in html and css, which is similar to flutters basic widgets, but there are external web component libraries that you can use for example that will give you some sort of design system. And you can choose depending on what you want to do. I think this is how it should be.
n7trd commentedon Apr 28, 2022
I'm also in favor of removing specific ui libraries (material, cupertino) out of the "core" framework.
*I don't see a "fragmented community" with this proposal just more choice and flexibility. If the material library should be the "de facto standard" for all platforms & use cases than its better to improve it and not bound it to the core framework. Also i'm not sure why cupertino is part of a cross-platform framework. Why should anyone use it on the web or desktop?!
iapicca commentedon Aug 24, 2022
copy paste from duplicate
Hixie commentedon Aug 25, 2022
The problem with unbundling the
materiallibrary is thatflutter createhas to create a template that uses a widget library of some sort, to help with the first hour experience (it's a lot more fun if you can just build an app without having to first figure out where to get buttons and so on). Most (all?) big platforms / UI frameworks / SDKs (e.g. the web, Windows, Android) do come with a baked-in set of widgets (actually Flutter is pretty unique in making its default set entirely optional and entirely defined in terms of lower-level primitives that you can just use directly; usually the default widgets are much more tightly integrated into the offering).We could depend on a split-out
materialinstead of bundling it with the SDK, but that would not really make any difference; you'd still need to pin the exact version for quality-control purposes, just like we do with vector_math. The effect would just be to slow development and make it harder for people to test their changes.materialdepends oncupertinowhich is why we bundlecupertinoas well.Responding to some comments above:
From @nikitadol:
You can use the
betachannel if you want faster updates. It uses the same release process asstable, but is updated monthly instead of every quarter. (You could also just usemaster, but that has had less testing.)As discussed above, the package and the SDK would be pinned in either case.
Not sure what this means.
Only one widget set needs to ship with Flutter. We could talk about changing to another built-in set, e.g. using
yaruas the out-of-the-box widget set, but I think in practice currently the material widgets are the most logical choice (since unlike most widget sets, they are designed to be cross-platform, like Flutter).@LasseRosenow:
If autocomplete is annoying, please file a bug specifically on that. We should definitely not be surfacing things in autocomplete that you don't want.
Unfortunately since we need to pin the material and flutter versions, this advantage is not available to us.
I agree that we should highlight other design languages. I'm not sure that this is more important than giving people a very fast out-of-box experience, though. Requiring that people think about what widget set they want when they're just experimenting with the framework is, IMHO, not conducive to a good experience.
@iapicca:
Having two packages would double the maintenance burden, which would just mean Material 2 no longer gets bug fixes.
I believe we could get rid of this now, we'd just need to move to handling the Material icons font like we do the Cupertino icons font. Please feel free to file a bug on this if it's important to you.
Actually the material.io team is already involved in
package:material's creation. It would not substantially change who was developing it if it moved out of the SDK. (Actually if it moved out of the Flutter project's purview into a less open-source-friendly project, it would probably mean that fewer people could contribute, which seems net worse.)In practice you can just not import package:material. We enforce a layering principle in the SDK to make sure nothing lower-level depends on the
materiallibrary.Responding also to some comments in #110195 from @alestiago:
As far as I can tell there's no mechanism to prevent people from depending on a
pubpackage any more than there is preventing them to depend onpackage:material. If you want to apply a style guide to disallow some package, it's relatively straight-forward to write a script to scan the code and flag uses of the banned package.As discussed above, even if we moved it into a separate package, it would still need to be pinned so that we could guarantee that the template app worked. So this would not solve that problem.
We already do this by enforcing layering such that nothing outside of
materialdepends onmaterial(and nothing outside ofcupertinoandmaterialdepends oncupertino). It would not make a meaningful difference to the layering if it was in a separate package.FWIW, I think there is a world where eventually we have
flutter createbuild from templates found in pub packages, and in that world we could consider having amaterialpackage that has a template and then the testing burden shifts sufficiently that we could think about not pinning the framework to the material library (or vice versa), by havingflutter createautomatically fetch thematerialpackage. It would need to be done very carefully. It would be interesting, though, since it would let us propose other such packages in the installation instructions (e.g. "flutter create -p macos" to get a macos-based template). (cc @christopherfujino)145 remaining items
redfieldchristabel commentedon Mar 27, 2026
with this decoupling, should i expect that the feature like default back button on IOS on scaffold widget will now gone?
Sshlale commentedon Mar 27, 2026
Piinks commentedon Mar 27, 2026
No, not at all. The libraries will work just as they do now. :)
redfieldchristabel commentedon Mar 28, 2026
how so, the scaffold widget are from widget whic is from flutter core lib rights. or scaffold are from material
Piinks commentedon Mar 30, 2026
The Scaffold class is defined in Material.
knotbin commentedon Apr 1, 2026
@Piinks is there any chance this tracking issue could be pinned on Flutter's issues page?
The sheer amount and nesting of subissues makes it hard to find and I'd say it's a pretty important and notable development that many people are very interested in.
flutter-triage-bot commentedon Aug 4, 2026
This issue is marked P1 but has had no recent status updates.
The P1 label indicates high-priority issues that are at the top of the work list. This is the highest priority level a bug can have if it isn't affecting a top-tier customer or breaking the build. Bugs marked P1 are generally actively being worked on unless the assignee is dealing with a P0 bug (or another P1 bug). Issues at this level should be resolved in a matter of months and should have monthly updates on GitHub.
Please consider where this bug really falls in our current priorities, and label it or assign it accordingly. This allows people to have a clearer picture of what work is actually planned. Thanks!
Piinks commentedon Aug 4, 2026
But there have been so many updates. 😅 Just not in the main umbrella issue. Sit tight, bot. Big doings are afoot.
tologonkudaiberdiuulu commentedon Nov 9, 2026
@Piinks any deadlines?
zigapovhe commentedon Nov 9, 2026
@tologonkudaiberdiuulu Deadlines for? material_ui and cupertino_ui have been 1.0 on pub.dev since August, so you can use them now. This one's still open because it's an umbrella. The old SDK libraries only got deprecated in #192957 and won't actually be removed until sometime next year (#172942), plus there are a few leftover sub-issues.