This is indeed a game-changing feature.
+1 (April)
Best,
Penguin
Bruno Cadonna <cado...@apache.org> 於 2025年4月2日 週三 下午3:43寫道：
> Great idea!
> PHP is the way to go!
>
> +1 (April)
>
> Best,
> Bruno
>
> On 02.04.25 02:18, Matthias J. Sax wrote:
> > Sounds like a great idea, Bill! We all know that Java is dead, but the
> > Web is here to stay!
> >
> > Combining the Web with AI sounds like a next natural step, to bring
> > KafkaStreams to the masses!
> >
> >
> > -Matthias
> >
> > On 4/1/25 11:58 AM, Bill Bejeck wrote:
> >> Dear Apache Kafka Community,
> >> I'd like to propose an exciting new extension to the Kafka ecosystem
> >> that I
> >> believe will revolutionize how we think about stream processing:
> >> KStreamsJS-PHP - a full PHP implementation of Kafka Streams with
> >> browser-based AI capabilities.
> >>
> >> Motivation
> >> As we all know, PHP is the backbone of the web. Despite numerous
> >> obituaries
> >> written for it since the early 2000s, PHP continues to power
> >> approximately
> >> 77.5% of all websites with a known server-side language. However, these
> >> millions of PHP developers have been unjustly excluded from the Kafka
> >> Streams ecosystem!
> >>
> >> Technical Overview
> >> KStreamsJS-PHP will be implemented as a PHP extension written
> >> primarily in
> >> JavaScript (transpiled to PHP) with a Rust core for performance. Key
> >> features include:
> >> 1. Full DSL API compatible with Kafka Streams but with PHP syntax:
> >>
> >> <?php
> >> $builder = new KStreamsBuilderJS();
> >> $kstream = $builder->stream("input-topic");
> >>
> >> $kstream
> >> ->filter(function($key, $value) { return $value->count > 1000;
> })
> >> ->map(function($key, $value) { return KeyValue::pair($key,
> >> strtoupper($value->name)); })
> >> ->to("output-topic");
> >> ?>
> >>
> >> 2. Browser-based State Stores: All state will be managed through
> >> IndexedDB
> >> in the user's browser. This dramatically simplifies operations as you no
> >> longer need to manage RocksDB instances.
> >> 3. AI-powered Stream Processing: The extension will include a
> >> revolutionary
> >> "AI Topology Builder" that watches as developers code and uses GPT-27
> >> (not
> >> yet released) to anticipate processing needs. The AI will automatically
> >> generate and inject stream processing topologies into your code as you
> >> think about implementing them!
> >> 4. WebWorker Threading: PHP, famously single-threaded, will leverage
> >> browser WebWorkers to achieve parallelism through a patent-pending
> >> technology we call "ClientThreading™️".
> >>
> >> Deployment Model
> >> The extension deploys PHP code to the browser via WebAssembly, with each
> >> user's browser becoming a node in your Kafka Streams application.
> Scaling
> >> is automatic - the more visitors to your website, the larger your
> >> processing cluster becomes!
> >>
> >> Project Timeline
> >> - Phase 1 (April 2025): Initial implementation and documentation
> >> - Phase 2 (June 2025): Integration with TikTok's short-form video API
> for
> >> visual stream processing
> >> - Phase 3 (September 2025): Launch of KafkaVerse - a metaverse
> >> environment
> >> where developers can physically walk through their streaming topologies
> >>
> >> Compatibility Guarantee
> >> KStreamsJS-PHP will maintain strict compatibility with Internet
> >> Explorer 6
> >> and Netscape Navigator.
> >>
> >> Security Considerations
> >> All stream processing happens client-side, so sensitive data never
> leaves
> >> the user's browser! As a bonus feature, users' CPU and RAM will be
> >> partially allocated to mining cryptocurrency to fund the project.
> >>
> >> Call for Feedback
> >> I'm seeking feedback from the community on this proposal, particularly
> >> regarding our innovative approach to AI-driven stream processing in the
> >> browser. We believe this project will cement Kafka's position in the
> >> "Web3
> >> Streaming" space.
> >>
> >> Please share your thoughts on this proposal. I'm particularly
> >> interested in
> >> how we can leverage newer browser APIs like the Ambient Light Sensor to
> >> adjust throughput based on the user's lighting conditions.
> >>
> >> Best regards,
> >> Elvis
> >>
> >
>
>
