wkhtmltopdf in Docker: How to fix common container errors in 2026

·1 views

The wkhtmltopdf project was archived in January 2023, with its last official packages released in May 2023 targeting Debian 12 and Ubuntu 22.04, leaving users on newer distros facing recurring errors. On Alpine Linux, the binary fails to run because it is compiled against glibc while Alpine uses musl, making the two incompatible with no reliable workaround. On Debian 13 or Ubuntu 24.04, older Dockerfiles that fetch the 0.12.6-1 release fail because that build depends on OpenSSL 1.1, which has been removed from modern distro archives. The recommended fix is to use the final 0.12.6.1-3 release on a Debian Bookworm or Ubuntu Jammy base image, as those builds were compiled against OpenSSL 3. Users are also warned that CVE-2022-35583, a server-side request forgery vulnerability, remains permanently unpatched, so untrusted HTML should never be passed to the tool.

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)

How Badge-Swipe Bonuses Turned Office Attendance Into a Game Employees Win

Companies including Google and Amazon have begun tying employee bonuses, performance reviews, and even termination decisions to office badge-swipe data, according to internal memos and a 2024 ResumeTemplates survey of 713 business leaders. The survey found that 18% of companies dock bonuses based on badge data, while 60% use it as their primary attendance-monitoring method. In response, workers have adopted 'coffee badging' — briefly swiping in, buying a coffee, and leaving — a practice that Owl Labs' 2025 report found 43% of hybrid workers were already doing. Data from Occuspace shows this behavior has inflated office occupancy figures by 15–20%, meaning compensation decisions are being made on inaccurate attendance records. Labor strategist John Frehse of Ankura warns that coffee badging signals deeper workplace dysfunction, and that companies are responding not by rethinking the metric but by adding more surveillance tools.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Windows NTFS Permissions Are More Complex Than Linux's Simple Permission Model

A developer completing cybersecurity training on HackTheBox found Windows permissions significantly more confusing than Linux's after finishing the Windows Fundamentals module. While Linux uses a straightforward read/write/execute model displayed via the 'ls -l' command, Windows relies on the 'icacls' tool and Access Control Lists (ACLs) made up of multiple Access Control Entries. Windows distinguishes between two ACL types: DACLs, which govern who can access a resource and how permissions propagate, and SACLs, which log access attempts. Basic Windows permissions include six categories such as Full Access, Read/Execute, and Modify, though the GUI and command-line tool display these differently, adding to the confusion. The repeated entries seen in icacls output are not duplicates but separate lines conveying permissions and inheritance rules for subfolders respectively.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Benchmark of 5 Managed Graph Databases Shows No Single Winner Across All Workloads

A developer benchmarked five managed graph databases — CognoDB Cloud, Neo4j AuraDB, Memgraph Cloud, FalkorDB Cloud, and ArangoDB Oasis — using the same dataset of 27,770 physics papers and 352,807 citation edges on free-tier cloud instances. Memgraph led on traversal queries and point lookups, completing 1-hop queries at a median 69ms, while Neo4j AuraDB finished second on those same tests. However, Neo4j reversed the standings on full-graph aggregation queries, outperforming Memgraph and leaving CognoDB and ArangoDB far behind at 1.8 and 4 seconds respectively. Under concurrent load testing with 10 and 40 simultaneous clients, four platforms scaled throughput roughly fourfold as expected, but ArangoDB's throughput barely moved, suggesting possible architectural or free-tier resource constraints. The results highlight that short-traversal performance and full-scan aggregation workloads stress graph databases in fundamentally different ways, making single-query benchmarks insufficient for real-world database selection.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Vector RAG Outperforms Classic RAG in Speed, Scale, and Cost Efficiency

Vector RAG (Retrieval-Augmented Generation) is emerging as the preferred retrieval method for production-grade large language model applications, according to a technical analysis published on DEV Community. Unlike classic RAG systems that rely on BM25 or TF-IDF text search, Vector RAG uses dense embeddings and vector indexes to cut query latency from 200–400 ms down to 30–80 ms. Approximate nearest neighbour algorithms used in libraries like FAISS, Milvus, and Pinecone reduce retrieval time by 5–10 times while maintaining recall above 95 percent. The approach also handles semantic similarity better, meaning misspellings, synonyms, and partial matches are less likely to cause hallucinations compared to classic retrieval methods. Vector indexes can be horizontally sharded and scaled with managed services, allowing teams to handle traffic spikes without rebuilding the entire index.

0 comments Read more at DEV Community