E-- Language Lets Developers Mix Plain English and Python Code Per Line

·1 views

E-- is a new programming language that allows developers to control how much natural English versus structured code appears in their programs, down to individual lines or values. It uses double-curly-brace placeholders where developers write English descriptions instead of explicit values, which a large language model resolves into concrete Python expressions at compile time. Once resolved, the compiler caches each slot's output, making subsequent builds fully deterministic and reproducible without further LLM calls. The language uses a closed set of English-style verbs and sentence structures that compile directly into standard Python. The design aims to make LLM involvement predictable and bounded, rather than a runtime variable that affects program behavior unpredictably.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Four Focus Management Mistakes Angular Developers Make and How to Fix Them

A developer guide published on DEV Community outlines how poor focus management in Angular apps creates serious barriers for keyboard and screen-reader users. Unlike sighted mouse users, keyboard navigators have a single focus position, meaning every UI change must deliberately direct where focus lands next. The article identifies four common failure points: SPA route changes that leave focus stranded, modals that trap focus but fail to return it to the trigger element on close, drawers and menus that open without shifting focus inside them, and destructive actions like row deletions that cause focus to disappear entirely. Recommended fixes include programmatically moving focus to page headings on navigation, using Angular CDK for dialog focus trapping and restoration, and leveraging afterNextRender to wait for DOM updates before redirecting focus. The guide emphasizes that automated accessibility audits cannot detect these focus-flow errors, making manual attention to focus discipline essential.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer Fixes Critical Flaws in AI Agent's Tamper-Evident Audit Log

A developer building an autonomous AI system discovered that its append-only audit log, meant to prevent history tampering, was fundamentally broken despite appearing functional. Two separate components were writing to the same log file in incompatible formats, with no locking mechanism, causing race conditions that corrupted the hash chain. Real corrupted records were found where stored hashes did not match their content, bearing the signature of concurrent writes. The team resolved the issues by routing all audit writes through a single locked implementation, anchoring the chain with an external cryptographic signature inaccessible to the runtime, and scheduling automated verification checks. The fixes highlight that a hash chain alone represents only a fraction of what a genuinely tamper-evident audit log requires.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How to Grant AWS Auditors Secure Access Using Cross-Account IAM Roles

Companies often grant external auditors AWS access by creating IAM users or identity provider accounts, but neither approach is necessary or secure. AWS supports a cross-account IAM role mechanism that allows auditors to assume a temporary role in your account without any credentials ever changing hands. The setup relies on four safeguards: a trust policy tied to the auditor's AWS account, an External ID condition to prevent unauthorized access, time-limited STS session credentials, and full CloudTrail logging of every action taken. Permissions are restricted to read-only managed policies, with optional billing access and a supplementary policy covering services like GuardDuty and Security Hub. A developer has published two CloudFormation templates on GitHub — one for a single account and one for an entire AWS Organization — each deployable in roughly five minutes.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How AI Decision Layers Can Replace Complex If-Else Logic in Automation Pipelines

A software developer has shared a practical pattern for integrating AI judgment calls into Python automation pipelines, addressing the limitations of traditional if/elif decision trees. The approach keeps deterministic tasks — such as HTTP calls, file I/O, and retries — handled by conventional scripts, while routing ambiguous decisions through small, tightly scoped AI prompts. A concrete example demonstrates a three-script pipeline enhanced with an AI triage step that classifies downloaded JSON files as PROCESS, REVIEW, or SKIP before further handling. The AI prompt is deliberately constrained to single-word outputs with enumerated choices, preventing ambiguous responses that could break downstream parsing. The author argues this separation of concerns — deterministic scripts for reliable tasks, AI for edge-case judgment — makes automation pipelines more robust without overcomplicating either component.

0 comments Read more at DEV Community