SShortSingh.
Back to feed

Bite's checker finds almost no gaps in Rich, Click and qs, but needed fixes to stay honest

0
·2 views

The author tested Bite, a tool that checks whether bug-fix changes are covered by tests, on 40 recent changes each in three well-known open-source projects: Rich, Click and qs. Bite noticed no untested changes in qs and only one small gap each in Rich and Click, which the author judged were not bugs. The initial run produced many false alarms on Rich, so Bite was updated to skip type-hint, version-specific and speed-only commits. A separate bug caused Click's old commits to load the current code, which the author fixed by giving each copy its own src/ folder.

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)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Static Windows driver vulnerability hunter flags BlackBoneDrv10.sys as critical

A developer built a static analysis tool that scans Windows .sys driver files for dangerous kernel APIs and risky IOCTL handlers without ever loading the driver. The tool parses import tables and strings, disassembles code to recover IOCTL codes, and assigns a severity score. When run on BlackBoneDrv10.sys it reported a CRITICAL rating, matching manual reverse engineering that found unvalidated arbitrary memory read/write. An initial false positive on the legitimate WinDivert64.sys was corrected by reweighting a commonly benign API, lowering it to LOW. The author stresses the results are triage signals to guide manual verification rather than proof of vulnerability.

0
ProgrammingDEV Community ·

AWS and Snowflake Combine Glue, S3, Iceberg and Cortex for Modern Data Pipelines

Organizations increasingly need scalable pipelines, governed access, open table formats, and AI-driven insights, and AWS and Snowflake can address these together without consolidating all data in one environment. The article describes a pattern where raw data is stored in Amazon S3, processed by AWS Glue, and loaded into Snowflake for SQL analytics and BI or AI applications. It highlights catalog federation, which lets AWS Glue access Snowflake Horizon Catalog metadata for Apache Iceberg tables, reducing metadata duplication and synchronization effort. The piece recommends secure, OAuth-based or secret-managed authentication over hardcoded credentials in production. It also notes an emerging integration of Amazon Quick with Snowflake Cortex AI via the Model Context Protocol to bring conversational analytics to business users.

0
ProgrammingDEV Community ·

Java JEPs That Reach Candidate Almost Always Ship, Census of 563 Proposals Finds

A pre-registered census by Munim, Inc. examined all 563 OpenJDK JDK Enhancement Proposals using the public JDK Bug System's change logs. It found that JEPs reaching Candidate status shipped at high rates, with 80.9% of those reaching it from 2018 to 2023 delivered within a year. Only 8.1% of JEPs that became Candidate between 2014 and 2020 failed to ship within five years, suggesting the main filter occurs before Candidate status. The study also found that 69.1% of the 68 proposals still in Draft or Submitted were over three years old, and that preview and incubator rounds made up a growing share of delivered features. The author notes that delivery dates were based on issue resolution dates, which often precede general availability by a median of 46 days.

0
ProgrammingDEV Community ·

SquadNote trial-participant feature stays open across Plane cycles until mobile checks and release are complete

A DEV Community article describes how a SquadNote feature ticket, SQN-30, was kept active when a weekly Plane cycle ended with mobile verification still outstanding. The author argues that marking the item Done just to close the cycle would misrepresent the unfinished checks. Plane documentation indicates unfinished work keeps its state and can move to another cycle, and a single work item can belong to both a cycle and a release module. The proposal separates development acceptance from production distribution by adding a distribution and post-release verification task alongside SQN-30. The workflow was built from existing acceptance criteria and verification records, with no Plane settings or states changed, and it was not operated over multiple weeks.