2025-12-08 Tech Briefing | Use prefetch to make pages load instantly, enhancing user experience.

Monday

Welcome to this edition, featuring 13 articles.

📚 Contents/📝 TL;DR

📚 Contents

📝 TL;DR

Primary

How to Make Pages Load Instantly Using link rel="prefetch":

Use prefetch to make pages load instantly, enhancing user experience.

How to Make Pages Load Instantly Using link rel="prefetch":

Use prefetch to make pages load instantly, enhancing user experience.

Data Engineering Weekly #247

Latest advancements in data and AI infrastructure from major companies

Data Engineering Weekly #247

Latest advancements in data and AI infrastructure from major companies

Snowflake Brings More Intelligent and Governed AI Technology for Enterprises | Technology Trends

Snowflake releases enterprise-level AI capabilities.

Snowflake Brings More Intelligent and Governed AI Technology for Enterprises | Technology Trends

Snowflake releases enterprise-level AI capabilities.

Reddit Migrated Its Comment Backend from Python to Go Microservices

Reddit successfully migrated its core comment backend from Python monolith to Go microservices.

Reddit Migrated Its Comment Backend from Python to Go Microservices

Reddit successfully migrated its core comment backend from Python monolith to Go microservices.

Has the Space AI Era Arrived? Musk Proposes Launching 1 Million Tons of AI Satellites Annually, Claims Lunar Factories Are the Cheapest Path to Compute Power

Musk proposes a space-based AI satellite concept.

Has the Space AI Era Arrived? Musk Proposes Launching 1 Million Tons of AI Satellites Annually, Claims Lunar Factories Are the Cheapest Path to Compute Power

Musk proposes a space-based AI satellite concept.

Essential

React CVE-2025-55182 Vulnerability Investigation and Remediation Guide

React/Next.js exposes a high-risk RCE vulnerability that requires immediate patching

React CVE-2025-55182 Vulnerability Investigation and Remediation Guide

React/Next.js exposes a high-risk RCE vulnerability that requires immediate patching

How Reddit Migrated Its Comment Functionality from Python to Go

Reddit successfully migrated its core comment functionality from Python monolith to Go microservices.

How Reddit Migrated Its Comment Functionality from Python to Go

Reddit successfully migrated its core comment functionality from Python monolith to Go microservices.

The Farewell Moment of 'Time Traveler' Werner Vogels

Werner Vogels' re:Invent farewell keynote speech.

The Farewell Moment of 'Time Traveler' Werner Vogels

Werner Vogels' re:Invent farewell keynote speech.

JavaScript Turns 30, Yet It Doesn't Own Its Own Name

JavaScript's 30th anniversary; the community is fighting to reclaim the trademark from Oracle.

JavaScript Turns 30, Yet It Doesn't Own Its Own Name

JavaScript's 30th anniversary; the community is fighting to reclaim the trademark from Oracle.

Global News Summary for December 7, 2025

Global news highlights for December 5-7, 2025: Chinese rockets, World Cup draw, Nobel Prize, US carrier, sunspots, US debt, India-Russia summit, China-France meeting, Seoul's first snow, Roman telescope, Indonesia's giant pandas, supermoon, Rockefeller Christmas tree, Russian president's visit to India.

Global News Summary for December 7, 2025

Google's Full Power Unleashed! Gemini 3 Deep Think Claims Multiple Reasoning SOTA, and the New Asian Team is Officially Announced

Google's Gemini 3 Deep Think model achieves SOTA in reasoning, introduces Workspace AI Agent, and announces a new team in Singapore.

Google's Full Power Unleashed! Gemini 3 Deep Think Claims Multiple Reasoning SOTA, and the New Asian Team is Officially Announced

Google's Gemini 3 Deep Think model achieves SOTA in reasoning, introduces Workspace AI Agent, and announces a new team in Singapore.

General

Event Loop: Mastering the Art of 'Slacking' Efficiently - A Guide to JS Asynchronous Programming

In-depth explanation of JS Event Loop, teaching the 'slacking' art of asynchronous programming

Event Loop: Mastering the Art of 'Slacking' Efficiently - A Guide to JS Asynchronous Programming

In-depth explanation of JS Event Loop, teaching the 'slacking' art of asynchronous programming

Vue Also Supports Declarative UI, Aligning with Mobile Kotlin and Swift, Abandoning HTML, OVSJS Upgrade, Check Out the New Syntax

Personal project attempting to introduce a new declarative UI syntax for Vue

Vue Also Supports Declarative UI, Aligning with Mobile Kotlin and Swift, Abandoning HTML, OVSJS Upgrade, Check Out the New Syntax

Personal project attempting to introduce a new declarative UI syntax for Vue

Primary

How to Make Pages Load Instantly Using link rel="prefetch":

Juejin Frontend Hottest•Published on December 3

Insight•Frontend Development•Score: 8/10

prefetchPerformance OptimizationUser ExperienceFrontendquicklink

🔬

Tech Insights

The article accurately identifies a core issue: optimizing only the first screen is insufficient. The 'consistency' and 'seamless feel' of the user experience are equally important. Using prefetch to load the next page resources without the user's awareness is a classic example of perceived performance optimization, which offers greater commercial value than mere technical metrics. It clearly distinguishes between preload and prefetch, avoiding a common confusion among beginners. It emphasizes the low priority nature of prefetch, which is the essence of utilizing 'wasted time'. The introduction of the quicklink library is very practical, providing an out-of-the-box automated solution that greatly reduces implementation barriers. Mentioning Next.js's built-in similar functionality confirms that this is a best practice in modern frameworks.

⚠️

Notable

The article does not delve deeply into the technical details or browser implementation principles of prefetch, such as DNS prefetch and preconnect. While it's a small trick, combining these techniques yields better results. It correctly emphasizes avoiding prefetching APIs, but if business scenarios allow (e.g., data is not sensitive and has low time sensitivity), prefetching some read-only, non-confidential, and cache-friendly APIs could be considered. However, this requires very careful architectural design and evaluation. In terms of mobile data-saving modes, navigator.connection.saveData is more of a signal, and in reality, browsers make intelligent decisions based on network conditions and resource priorities. Our manual intervention space may not be as significant as we imagine.

📈

Perspectives

In the current environment of high traffic costs and scarce user attention, any frontend optimization that can increase conversion rates and reduce bounce rates is a rare resource. prefetch optimization, which does not require changing business code, has a very high ROI and is a classic case of achieving much with little effort. The domestic frontend market is highly competitive, and performance optimization has become not just an add-on but a baseline. Everyone is striving for the ultimate user experience, especially in content consumption, e-commerce, and SaaS sectors. Even minor loading delays can directly impact user retention and commercial revenue. Teams that can perfect these 'small details' often achieve a significant edge in user experience. Do not underestimate these 'experience debts'; they accumulate into user loss.

Source

◆

Data Engineering Weekly #247

Data Eng Weekly•Published on December 8

Insight•AI and Frontier Technologies•Score: 8/10

Data EngineeringAI GovernanceML PlatformReal-time DataRust

🔬

Tech Insights

In my view, there are several key points worth attention: 1. DoorDash's collaborative AI ecosystem: From single-agent systems to multi-agent collaboration, and further to shared memory and clustered A2A (Agent-to-Agent) collaboration, this evolutionary architecture clearly demonstrates the path from simple to complex AI applications. Particularly, the unified platform's hybrid search and Schema-aware SQL generation are the key infrastructure that truly enables AI to 'do work'. 2. LinkedIn's Venice and FishDB: Venice's ingestion pipeline optimization and FishDB's reimplementation of the feed recommendation engine in Rust both point to extreme performance optimization and cost control. Especially FishDB, replacing JVM with Rust directly cut hardware costs by 50%, and kept p99 latency under 40ms. This is a cost-effective model that many domestic companies would dream of. The penetration of Rust in high-performance infrastructure is already a trend. 3. Dropbox's Context Engineering: AI agents often face the issue of 'overload' in long-running and complex scenarios. Dropbox improved inference quality and efficiency by using a unified search tool, knowledge graph filtering, and specialized agent delegation. This reveals one of the core bottlenecks in deploying AI agents, which is how to efficiently and accurately manage context. 4. Lyft's hybrid ML platform architecture: LyftLearn moved from a pure Kubernetes setup to a Kubernetes + SageMaker hybrid architecture. This reflects the limitations of a single platform in ultra-large-scale ML training scenarios. The adoption of hybrid cloud and multi-cloud strategies in the ML field aims to balance flexibility, cost, and specialized service capabilities.

📈

Perspectives

From these major companies' moves, several trends in the domestic data and AI field are becoming increasingly clear: 1. Heavy investment in AI infrastructure: From agent ecosystems to the integration of foundation models, domestic companies have already made significant investments at the AI model level. However, the challenge remains in efficiently, stably, and controllably integrating these models into existing business operations. DoorDash's collaborative AI and Dropbox's Context Engineering provide good insights, indicating that AI engineering and governance will be the focus in the next phase. 2. Real-time data and performance optimization: LinkedIn's case once again proves that for massive data and high-concurrency scenarios, performance optimization is never-ending. The use of high-performance languages like Rust in infrastructure layers will grow. Domestic teams still hesitating between Java and Go should seriously consider Rust's strategic position. 3. MLOps and platform evolution: Lyft's hybrid ML platform architecture suggests that many domestic companies building their own ML platforms may also move towards hybrid or multi-cloud solutions to balance development efficiency, resource costs, and elasticity. Building a fully self-developed platform comparable to AWS SageMaker requires significant investment, and trade-offs are an art. 4. Data quality and contracts: Grab's real-time data quality monitoring emphasizes the importance of data contracts. In the era of data-driven decision-making, data quality is the foundation. However, the popularization of data contracts in China still needs to be further improved, which can effectively prevent many data incidents.

Source

◆

Snowflake Brings More Intelligent and Governed AI Technology for Enterprises | Technology Trends

InfoQ - AI & LLM•Published on December 8

Insight•AI and Frontier Technologies•Score: 8/10

SnowflakeEnterprise AIData AgentsCortex AI SQLMLOps

🔬

Tech Insights

In my view, Snowflake's latest update has several highlights worth noting: 1. Snowflake Intelligence: This is an enterprise-level agent that breaks down data silos through natural language queries and automates the semantic layer. Essentially, it is driving the democratization of data and insights, allowing non-technical personnel to directly extract value from data, which is something many enterprises aspire to. 2. Cortex Agents: Emphasizes building trustworthy data agents that can coordinate information across both structured and unstructured data sources and support mainstream LLMs. The key lies in its deep integration of AI capabilities with data governance, ensuring data security and contextual accuracy. 3. Cortex AI SQL: Builds multi-modal AI pipelines using familiar SQL language, claiming performance improvements of 3 to 7 times and built-in cost governance. This is essentially a game-changer in lowering the AI application threshold, enabling traditional data engineers to easily leverage AI and significantly improving development efficiency and cost control. 4. Enhanced MLOps: Integrates experiment tracking, online feature storage, and low-latency prediction services. This means the entire model development to production deployment pipeline is completed within a single platform, avoiding the complexity and risks of data leaving the system.

⚠️

Notable

Although Snowflake's vision is grand, there are several drawbacks to consider: 1. The claim of "building enterprise-level agents without technical background" seems to me to be more of a marketing phrase. In reality, enterprise AI projects require deep expertise in requirement definition, data preparation, model selection, and tuning. Relying on "zero-code" to solve all problems is unrealistic. 2. The "3 to 7 times performance improvement" claimed by Cortex AISQL lacks third-party independent verification, we should be cautious of this marketing exaggeration. Actual performance often depends on specific scenarios and data complexity. 3. For domestic enterprises, data compliance and localized service capabilities remain significant challenges. Data not leaving the country is a red line, and whether Snowflake's global architecture can meet the stringent regulatory requirements of China remains to be seen.

📈

Perspectives

In my opinion, Snowflake's move is a wise one, directly targeting the enterprise AI market. It aims to capture market share from traditional AI/ML platforms (such as Databricks and SageMaker) by adopting a data platform + AI capabilities integration model. Its strengths lie in data-native capabilities, governance, and cost control, which can be attractive to Chinese enterprises with large data volumes and a focus on compliance. However, domestic cloud providers like Alibaba Cloud and Tencent Cloud are also actively developing similar capabilities, and Snowflake faces challenges in localization and ecosystem building. In the long run, the deep integration of data and AI is an irreversible trend. Whoever can provide a more seamless, compliant, and context-appropriate solution will succeed in the Chinese market. Now in December 2025, Chinese enterprises are continuously increasing their AI investments, and the importance of data governance and cost control is growing. Whether Snowflake can find its niche in the Chinese market remains to be seen.

Source

◆

Reddit Migrated Its Comment Backend from Python to Go Microservices

InfoQ - Architecture•Published on December 7

News•Software Architecture and Engineering Practices•Score: 7/10

RedditMicroservices MigrationPythonGoPerformance Optimization

Source

◆

Has the Space AI Era Arrived? Musk Proposes Launching 1 Million Tons of AI Satellites Annually, Claims Lunar Factories Are the Cheapest Path to Compute Power

InfoQ - AI & LLM•Published on December 8

News•AI and Frontier Technologies•Score: 7/10

MuskSpace AISatellite Compute PowerLunar FactoryKardashev Civilization

🔬

Tech Insights

Musk's proposal, although absurd, has its own logic: 1. Space AI Satellites: He believes that localized AI computing satellites in low-latency sun-synchronous orbits will become the most cost-effective AI compute power method within three years. This points toward the extreme form of distributed and edge computing. 2. Lunar Factory: A more advanced plan is to build a satellite factory on the Moon, using electromagnetic railguns to accelerate satellites, achieving 100 terawatts of compute power, moving toward a Kardashev Type II civilization. This is not just technology; it's an imagination of human civilization's evolution. 3. Compute Bottleneck: He argues that Earth's available power resources are scarce, and space offers a faster path for scalable expansion. This indeed hits the current pain points of energy and cooling in AI development.

⚠️

Notable

Musk's remarks have more drawbacks than merits, and they are truly 'unconventional and shocking': 1. Annual launch of 1 million tons of satellites: This is not just a cost issue, but a question of engineering and resource feasibility. Current rocket launch capacity and manufacturing costs are completely unable to support such a massive scale. This is not a proposal that can be discussed in the short term. 2. Lunar Factory: This is purely a sci-fi concept. In the foreseeable decades, even longer, it has no practical feasibility. It seems more like a 'ultimate goal' to attract attention. 3. Equating 'compute power' simply with 'watts' and 'tons' is overly simplistic. AI compute power is not only about hardware power, but also involves complex software stacks, algorithm optimization, data transmission, and storage, which ignores the complexity of the AI ecosystem. 4. This appears more like a 'marketing stunt' than a rigorous technical roadmap. Musk is well aware of this, using extreme proposals to spark discussion and maintain his status as a 'thought leader' in the tech community.

📈

Perspectives

Although Musk's statements are exaggerated, they indeed reflect the global anxiety about the limits of AI compute power. As of December 2025, the cost, energy consumption, and cooling challenges of AI compute power on Earth are becoming increasingly prominent, prompting the industry to seek new solutions. In the short term, Musk's 'space compute' proposal will not affect the existing AI compute power market structure. It is more of a distant vision. The domestic AI compute power market still focuses on chip domestication, liquid-cooled data centers, and compute power scheduling optimization. Musk's vision is too far from the Chinese market, serving more as a topic for casual conversation. However, it reminds us that the compute bottleneck will eventually arrive, and we need to prepare in advance.

Source

Essential

React CVE-2025-55182 Vulnerability Investigation and Remediation Guide

Juejin Frontend Hottest•Published on December 4

News•Frontend Development•Score: 10/10

ReactCVERCEsecurity vulnerabilityNext.js

🔬

Tech Insights

This vulnerability is a major event in the frontend world, and it's essential to raise the alarm and start remediation immediately: 1. RCE, unauthenticated, CVSS 10.0: These three keywords together indicate the highest level of threat. Attackers can execute arbitrary code on the server with no authentication, essentially opening the server's door. It specifically highlights Next.js, which is a major hotspot. 2. RSC Flight protocol deserialization: The root cause lies in the insecure deserialization handling of the Flight protocol in React Server Components (RSC). This serves as a reminder that any data involving network transmission and deserialization must undergo extremely strict validation and sandbox isolation. As RSC is a core feature of React 19, its underlying security issues have a massive impact. 3. Detailed investigation and remediation steps: The article offers very clear and actionable steps, including checking dependencies, upgrading core packages, upgrading frameworks (especially with a detailed version mapping for Next.js), dependency scanning, and subsequent monitoring and protection recommendations. This is a life-saving guide for teams managing production environments.

⚠️

Notable

Honestly, a CVSS 10.0 RCE vulnerability, especially one that can be exploited with default configurations, is an unacceptable mistake for React and Next.js—two widely used frameworks. This exposes several deeper issues: 1. Inadequate security design: RSC, as a core feature, has a significant security flaw in its Flight protocol deserialization mechanism. This suggests that security considerations may not have been given sufficient priority during the design phase, or that risk assessments were severely lacking. 2. Insufficient test coverage: Such a severe vulnerability exists across multiple versions of React 19, indicating potential blind spots in pre-release security testing, penetration testing, and coverage of the deserialization attack surface. 3. Supply chain security risks: This core library vulnerability directly affects countless downstream applications. It reminds us once again that even top-tier open-source projects like React are not immune to security issues. Continuous auditing of third-party dependencies and a rapid response mechanism are crucial.

Source

◆

How Reddit Migrated Its Comment Functionality from Python to Go

ByteByteGo•Published on December 7

Insight•Software Architecture and Engineering Practices•Score: 9/10

RedditMicroservices MigrationPythonGoData ConsistencyTap CompareSister Datastores

Source

◆

The Farewell Moment of 'Time Traveler' Werner Vogels

InfoQ - AI & LLM•Published on December 8

Insight•Software Architecture and Engineering Practice•Score: 9/10

Werner Vogelsre:InventDesign for FailureServerlessRenaissance DeveloperBuilders Spirit

🔬

Tech Insights

Werner's speeches have never been about chasing trends, but rather pointing to the essence of technology and engineering. This 'final lesson' is particularly valuable: 1. 'Design for Failure' (Design for Failure): This is the cornerstone of AWS's success and the 'Bible' of distributed system architecture. In the AI era, the emphasis on spec-driven development, replacing ambiguity in natural language with standardized specifications, aligns with the philosophy of 'Design for Failure'. This is a demonstration of engineering discipline. 2. From Serverless to Codeless: Werner has always been committed to 'developer liberation', from Serverless to the current generation of AI-driven programming (Codeless), all aim to hide infrastructure complexity and reduce technical debt. This shows that simplifying the development experience is an eternal pursuit. 3. The five key traits of the 'Renaissance Developer': maintaining curiosity, systems thinking, effective communication, ownership mentality, and becoming a polymath. This methodology is not only a personal growth path, but also a thinking framework for top-tier architects, pointing to the core competitiveness of future technical talents. 4. Builders Spirit: 'Doing the right thing where no one is watching'. This emphasizes the responsibility and values of technical professionals, which are more important than mere technical ability. The cases of Andon Cord and durability reviews exemplify how this spirit can be institutionalized, making it a valuable lesson for all teams.

⚠️

Notable

Although Werner's speech is filled with wisdom, as someone within the industry, I have some 'cynical' opinions: 1. The article praises Werner excessively, although he is indeed great, but lacks some neutral and critical perspectives. Over-glorifying individuals may cause people to overlook the power of teams and the times they live in. 2. The concept of the 'Renaissance Developer' is admirable, but in the 'screwdriver' work environment prevalent in domestic large companies, it is extremely difficult to implement. Often, individual development is constrained by organizational structures and KPI evaluations, making it hard to truly embody 'being a polymath' and 'ownership mentality'. 3. Spec-driven development is proposed in the AI era, but the rapid iteration and uncertainty of AI make strict spec-driven development too rigid, potentially slowing down innovation. How to balance specifications and agility is a new challenge in the AI era.

📈

Perspectives

Werner Vogels' farewell marks the end of an era for AWS, but his profound ideas will continue to influence cloud computing and software engineering. His concepts, such as 'Design for Failure' and 'Serverless', have become industry standards, and domestic cloud vendors and enterprise architects are learning and practicing them. Especially at this time point of December 2025, when domestic companies generally face cost reduction and efficiency improvement pressures, Werner's emphasis on 'Builders Spirit' and the traits of the 'Renaissance Developer' provides strong reference value for Chinese technical talent cultivation and team culture building. These core values are more important than blindly chasing short-term technological trends; they help teams establish a long-term perspective, building more robust and sustainable systems. After all, technology serves people, and 'doing the right thing' is the highest honor for technical professionals.

Source

◆

JavaScript Turns 30, Yet It Doesn't Own Its Own Name

InfoQ - Cloud•Published on December 7

Insight•Frontend Development•Score: 8/10

JavaScriptTrademark DisputeOracleECMAScriptNode.jsFrontend Ecosystem

Source

◆

Global News Summary for December 7, 2025

Global Vision•Published on December 7

News•Comprehensive News•Score: 8/10

International NewsTechnology AdvancesSports EventsEconomic AnalysisGeopolitics

📝

Summary

This article summarizes key global events from December 5 to 7, 2025, spanning technology, sports, politics, economics, and natural phenomena. It includes the launch of China's Changzheng-8A rocket with internet satellites, progress on the 'Guowang Constellation', results of the 2026 World Cup draw, Nobel Prize activities in December 2025, the US Ford-class carrier's ongoing deployment in the Caribbean, expanding sunspot groups and solar storm warnings, US debt surpassing $3 trillion, the 23rd India-Russia summit and trade agreements, France's President Macron concluding his visit to China and the China-France joint declaration, Seoul's first snow of the season, completion of the Roman space telescope, naming of a new panda cub in Indonesia, the last supermoon of 2025, the lighting of the Rockefeller Center Christmas tree in New York, and the Russian president's visit to India.

🔬

Tech Insights

Global Technology Breakthroughs: China's Changzheng-8A rocket, which first uses coal-based aviation kerosene, continues to advance the 'Guowang Constellation' in the low-orbit internet satellite field, directly influencing the future global communication market landscape. Meanwhile, the completion of NASA's Roman space telescope signals a significant enhancement in deep space exploration capabilities. Geopolitical Dynamics: The India-Russia summit signed a trade agreement aiming for $100 billion bilateral trade by 2030 with a 96% rate of local currency settlement, indicating a deepening economic integration between the two countries. The informal meeting between China and France's leaders, along with their joint declaration, highlights shared consensus and cooperation intentions in global governance and multilateral trade systems. Macroeconomic Risks: US debt has first surpassed $3 trillion, and the IMF predicts it will reach 143.4% of GDP by 2030, introducing long-term uncertainty to the global economy. This poses potential debt crisis risks for enterprises and countries reliant on the US dollar system.

📈

Perspectives

Energy Transition: The use of coal-based aviation kerosene by Changzheng-8A signifies China's active exploration in aerospace fuel cost reduction and efficiency as well as environmental sustainability. This could lead to the formation of new technical barriers and competitive advantages in the commercial aerospace sector. De-dollarization Acceleration: The near 100% rate of local currency settlement between India and Russia, along with their $1 trillion trade goal, clearly reflects emerging economies' resistance to US dollar hegemony. This provides crucial insights and practical experience for Chinese enterprises in international trade settlement strategies. Global Economic Downturn: US debt reaching a new high is not just an issue for the US but could also affect the global economy through spillover effects. Chinese enterprises must be more cautious in assessing overseas investment risks and focus more on supply chain resilience.

Source

◆

Google's Full Power Unleashed! Gemini 3 Deep Think Claims Multiple Reasoning SOTA, and the New Asian Team is Officially Announced

InfoQ - AI & LLM•Published on December 8

Insight•AI & Frontier Technologies•Score: 8/10

KubeConAI ComputingKubernetesRayPyTorchvLLMLLMOpsDistributed AI

📈

Perspectives

In 2025, AI computing is no longer just about model training; it involves complex data processing, training, and inference pipelines, presenting engineering challenges. Teams still running models on single machines are ready to embrace distributed AI computing stacks. The combination of Kubernetes and Ray is essentially the golden pair for building large-scale AI infrastructure. Any team aiming to make an impact in AI must deeply understand and master the synergy between these two, otherwise, even the best AI models will not run or perform well. The market's demand for AI platforms has evolved from 'can run' to 'efficient, stable, observable, and governable', indicating that investments in LLMOps and MLOps will continue to grow, becoming the main battlefield for future technical investments.

Source

General

Event Loop: Mastering the Art of 'Slacking' Efficiently - A Guide to JS Asynchronous Programming

Juejin Frontend Hottest•Published on December 6

Insight•Frontend Development•Score: 6/10

JavaScriptEvent Loopasynchronous programmingmicrotasksmacrotasks

🔬

Tech Insights

The article uses the 'JS worker slacking' analogy to vividly explain the mutual exclusivity between processes, threads, JS engine threads, and rendering threads, as well as the 'helplessness' and 'wisdom' of JS single-threaded task processing. This down-to-earth explanation is more accessible than official documentation. The 'four-step slacking mantra' for Event Loop is summarized very effectively: 'synchronous tasks first', then 'clear microtask queue', followed by 'rendering the page', and finally 'execute a task from the macrotask queue'. This provides a clear execution model. Through multiple code examples (setTimeout, Promise.then, async/await), the article thoroughly breaks down the execution order at each step, especially the explanation of code entering the microtask queue after await, which hits a common misunderstanding point.

Source

◆

Vue Also Supports Declarative UI, Aligning with Mobile Kotlin and Swift, Abandoning HTML, OVSJS Upgrade, Check Out the New Syntax

Juejin Frontend Hottest•Published on December 6

Insight•Frontend Development•Score: 6/10

VueDeclarative UIFrontend FrameworkOVSJSDSL

🔬

Tech Insights

In my opinion, this project has several notable highlights: 1. Introduction of Declarative UI: The author attempts to bring the concept of declarative UI from mobile platforms to the web frontend, which itself is a valuable exploration direction. Avoiding XML/HTML tag nesting and using code structure to directly express UI can theoretically improve development efficiency and readability, especially in scenarios with high componentization and complex logic. 2. Integration of UI and Logic: The author's emphasis on 'supporting code writing in declarative UI' and 'not rendering code blocks' does indeed bring UI and logic closer together. For components where business logic is tightly bound to the UI, this may offer convenience. 3. Self-developed Parser: To implement the new syntax, the author developed their own parser, showcasing a certain technical depth and implementation capability. By converting the new syntax into Vue's rendering function through AST transformation, it ensures compatibility with the Vue ecosystem, which is far more complex than simple syntactic sugar.

⚠️

Notable

To be honest, this approach is still far from 'abandoning HTML', and it has a clickbait feel. Declarative UI in web frontend has already been well addressed by Vue and React's JSX/Template. The biggest issue with this 'reinventing the wheel' syntax is: 1. Ecosystem and Learning Curve: Introducing a completely new DSL (Domain-Specific Language) means developers must learn a new syntax, and there's a lack of mature toolchains, community support, and a large number of third-party components. This poses a high risk for teams that prioritize efficiency and stability. 2. Maintainability and Compatibility: Although the author mentions 'UI and logic being closer together', excessive integration could result in unclear code responsibilities and difficulty in separation. Moreover, the Vue official team has not adopted this approach, so long-term maintenance and future upgrades are major uncertainties. 3. Uncertain Benefits: Compared to existing JSX or Vue Template, this syntax similar to Kotlin DSL may not bring significant efficiency improvements in actual development. It requires extensive practical validation. Often, the complaints are not about HTML itself, but rather the complexity of JS frameworks.

📈

Perspectives

The domestic frontend community is highly sensitive to 'new syntax' and 'new paradigms', but most are just noise without substance. Mobile declarative UI succeeded because they built their ecosystems from scratch and had strong official support. The web frontend is already an extremely mature and large ecosystem, with HTML/CSS/JS as its foundation. Any attempt to 'abandon' these technologies faces significant inertia resistance. I believe this personal project mainly provides a technical idea and inspiration, proving that there is still room for innovation in the syntax layer of web frontend. However, for it to be truly applied in production environments, it would need strong backing from the Vue official team. In China, the focus is more on how to improve efficiency within existing frameworks through engineering practices, component libraries, and best practices, rather than introducing a new, high-uncertainty syntax layer.