Aura package portfolio review: archive / modernize / keep

[koriym]

Aura has been around about 15 years. Some packages are clearly stale, some are still releasing this year. Before deciding what to archive, I want to agree on the question we're actually asking.

The question isn't "is this old?" It's "should we carry this for another 15 years?"

In the AI-coding era, modernization cost has dropped a lot. PHP version bumps, CI, types, docs — much cheaper than before. So age alone is no longer a reason to archive. Value is.

Framework line

Aura.Framework already says it's unmaintained. Aura.SqlMapper_Bundle says "use Atlas." system is tagged deprecated. These three can be archived immediately.

The framework story as a whole looks over. If we agree on that, the　kernel/project/installer family follows: Framework_Demo, Framework_Project, Cli_Project, Web_Project, Cli_Kernel, Web_Kernel, Project_Kernel, installer-default, installer-system, Example.Testing.

Aura as decoupled libraries, on the other hand, may still be worth carrying.

Proposed classification

Four buckets:

1. Active / Modernize — still solves a real problem, keep investing
2. Revive / Redesign — good idea, API needs a modern rewrite
3. Legacy / Maintenance-only — works for existing users, not recommended for new projects
4. Archive / Superseded — PSR or successor already covers it

Draft assignment

Active / Modernize

Aura.Di, Aura.Sql, Aura.SqlQuery, Aura.SqlSchema, Aura.Router, Aura.Session, Aura.Accept, Aura.Html, Aura.Marshal

(Several of these had releases in 2025–2026.)

Revive / Redesign

Aura.Filter (+ Filter_Interface), Aura.Auth, Aura.Intl

Legacy / Maintenance-only

Aura.Web, Aura.Payload (+ Payload_Interface), Aura.Input

Archive / Superseded

Standards / successors took over:

• Aura.Autoload → Composer
• Aura.Http → PSR-7/17/18
• Aura.Signal → PSR-14
• Aura.Dispatcher → existing invoker libraries
• Aura.View → Qiq (author's own successor)
• Aura.SqlMapper_Bundle → Atlas (author's own successor)
• Aura.Cli → mature CLI libraries in the wider ecosystem

Framework line: see above.

Next steps

1. Agree on the four-bucket model.
2. Archive the already-deprecated three first (Aura.Framework, Aura.SqlMapper_Bundle, system).
3. Decide each remaining repo's bucket.
4. Add archive notices with migration pointers.
5. Update Packagist + the website to reflect status.

The goal isn't fewer repos. It's a clearer story about what Aura is in 2026.

[harikt]

• Agree on the four-bucket model.
• Archive the already-deprecated three first (Aura.Framework, Aura.SqlMapper_Bundle, system).
• Decide each remaining repo's bucket.
• Add archive notices with migration pointers.
• Update Packagist + the website to reflect status.

Legacy

• Aura.Web : I know https://github.com/sapienphp/sapien exists.

Revive / Redesign

To the list, I will also add

• Aura.Input
• Aura.Dispatcher
• Aura.View : Qiq is very different, so we can still maintain this package.

I love Aura.Signal, but I know it is not used that much.

Another alternative for choosing the package is using packagist and looking at the dependents : https://packagist.org/packages/aura/view/dependents?order_by=downloads

[koriym]

Right, let’s keep these.

• Aura.Input
• Aura.Dispatcher
• Aura.View
• Aura.Signal

[pmjones]

Thanks @koriym and @harikt for producing this!

• Aura.Http might also point to HTTPlug and the Symfony HTTP Client.

• Aura.Web might point either to Sapien, or (preferably) implement the newer Star-Interop Request and Response interfaces.

Aura.View pointing to Qiq would be nice, but Qiq combines a variation of Aura.Html, so pointing to Qiq might be more than people expect from a replacement.

Those are the things that come to mind right now; when there is a full draft plan in one place I'd be happy to look it over with you all again.

[harikt]

Active / Modernize

• Aura.Sql
• Aura.SqlQuery
• Aura.SqlSchema
• Aura.Accept
• Aura.Di
• Aura.Dispatcher
• Aura.Intl
• Aura.Marshal
• Aura.Router
• Aura.Payload_Interface
• Aura.Payload

Revive / Redesign

• Aura.Session_Interface : New repo need to be created with help from @pmjones .
• Aura.Session
• Aura.Auth
• Aura.Filter_Interface
• Aura.Filter
• Aura.Input

For the Input and Filter, we need Html. I have recently fixed some of the issues auraphp/Aura.Html#69 . If qiq do support it, I will be more happy to remove Aura.Html and Aura.View from this list. Else we may want to maintain.

• Aura.Html → Qiq (author's own successor)
• Aura.View → Qiq (author's own successor)

Legacy / Maintenance-only

• Aura.Web -> Implement the newer Star-Interop Request and Response interfaces ? If Sapien is going to implement this, then we don't need to implement here. But I can see ~ 200 install per month . https://packagist.org/packages/aura/web/stats . So maintinance only for this ?

Archive / Superseded

• Aura.Autoload → Composer
• Aura.Http → HTTPlug and the Symfony HTTP Client or https://packagist.org/providers/psr/http-client-implementation .
• Aura.Signal → PSR-14 may be point to https://packagist.org/providers/psr/event-dispatcher-implementation
• Aura.SqlMapper_Bundle → Atlas (author's own successor)
• Aura.Cli → climate, symfony console

Do you guys agree with above ?

[koriym]

After looking at the install stats, I'd like to propose a slightly different framing.

The Active vs Legacy distinction matters more than Archive.

Almost every package still has a meaningful install base. As I noted in the opening, AI can now handle PHP version bumps and CI updates mechanically at near-zero human cost — so Legacy doesn't have to mean abandoned. It can mean: "kept alive automatically, no new features, no human maintenance burden."

With that in mind:

• Archive: only the framework line (already deprecated) + aura/http (49 installs/month, PSR-7/18 is a complete replacement)
• Legacy / AI-maintained: everything else with lower activity — PHP version support and CI kept current by AI, no feature work
• Active / Modernize: packages with clear ongoing demand and development

Also, a correction on the stats: @harikt's "~200/month" for Aura.Web is the daily average, not monthly — actual monthly installs are ~7,300. And aura/autoload is trending upward since 2020, worth investigating before any archive decision.

30-day install counts

Package	Last 30 days
aura/intl	71,911
aura/sql	53,026
aura/sqlquery	46,519
aura/router	21,738
aura/cli	19,878
aura/session	13,711
aura/autoload	11,241
aura/web	7,347
aura/di	6,861
aura/filter	4,577
aura/html	3,493
aura/payload	3,335
aura/accept	2,495
aura/sqlschema	2,432
aura/view	2,053
aura/auth	1,410
aura/input	1,170
aura/dispatcher	895
aura/signal	623
aura/marshal	388
aura/http	49

[koriym]

Aura PHP CI Support Versions

Package	CI	PHP Versions	Last 30 Days Installs
Aura.Intl	GH	5.6, 7.0, 7.1, 7.2, 7.3, 7.4, 8.0, 8.1, 8.2	80,670
Aura.Sql	GH	8.4, 8.5	55,779
Aura.SqlQuery	GH	5.6, 7.0, 7.1, 7.2, 7.3, 7.4, 8.0, 8.1, 8.2, 8.3, 8.4, 8.5	48,845
Aura.Router	GH	5.5, 5.6, 7.0, 7.1, 7.2, 7.3, 7.4, 8.0, 8.1, 8.2, 8.3, 8.4	23,725
Aura.Cli	GH	7.2, 7.3, 7.4, 8.0, 8.1, 8.2	20,634
Aura.Session	GH	8.1, 8.2, 8.3, 8.4, 8.5	14,078
Aura.Autoload	Travis	5.3, 5.4, 5.5, 5.6, hhvm, 7	12,867
Aura.Di	GH	8.0, 8.1, 8.2, 8.3, 8.4, 8.5	7,637
Aura.Web	GH	5.4, 5.5, 5.6, 7.0, 7.1, 7.2, 7.3, 7.4, 8.0, 8.1	7,425
Aura.Filter	GH	8.4, 8.5	4,615
Aura.Html	GH	8.4, 8.5	3,979
Aura.Payload	GH	5.6, 7.0, 7.1, 7.2, 7.3, 7.4, 8.0, 8.1, 8.2	3,330
Aura.Accept	GH	7.2, 7.3, 7.4, 8.0, 8.1, 8.2	2,873
Aura.SqlSchema	Travis	5.3, 5.4, 5.5, 5.6, hhvm, 7	2,506
Aura.View	GH	5.4, 5.5, 5.6, 7.0, 7.1, 7.2, 7.3, 7.4, 8.0, 8.1	2,484
Aura.Auth	GH	7.2, 7.3, 7.4, 8.0, 8.1	1,467
Aura.Dispatcher	Travis	5.4, 5.5, 5.6, hhvm, 7	903
Aura.Marshal	GH	7.2, 7.3, 7.4, 8.0, 8.1, 8.2, 8.3, 8.4, 8.5	441
Aura.Includer	Travis	5.3, 5.4, 5.5, 5.6, hhvm, 7	4