Episodes

  • Keep Cypress, keep Selenium, or rewrite in Playwright? | The Platform Radar
    Nov 12 2026

    A Node upgrade just put your end-to-end suite back on the table, and a popularity chart is the worst way to settle it. This deep dive checks what your suite actually needs against the limits each tool documents about itself.


    You'll learn which of your suite's hard needs decide this: the language your tests are written in, WebKit coverage, multiple tabs and origins, a matrix of branded browsers across versions, and a parallel grid. We check each one against the limits that Playwright, Cypress and Selenium document themselves. We also look at what a developer survey can and can't tell you, and end with the checks to run on Monday so you can decide in one sitting.


    Chapters

    00:00 Intro

    01:30 Why your Node bump reopens this

    04:26 Where Playwright fits

    07:58 When keeping Cypress is right

    11:33 When keeping Selenium is right

    15:14 What the survey can't tell you

    18:24 What to do on Monday


    Sources (retrieved 2026-11-09)

    - Stack Overflow Developer Survey 2026 – Code testing tools (usage), Stack Overflow (2026-10-06): https://survey.stackoverflow.co/2026/technology/data/code-test

    - Stack Overflow Developer Survey 2026 – Code testing tools (admired and desired), Stack Overflow (2026-10-06): https://survey.stackoverflow.co/2026/technology/data/da-code-test

    - The results of the 2026 Developer Survey are here, Stack Overflow (2026-10-06): https://stackoverflow.blog/2026/10/06/the-results-of-the-2026-developer-survey-are-here/

    - Trade-offs, Cypress.io (2026-11-09): https://docs.cypress.io/app/references/trade-offs

    - Cross Browser Testing, Cypress.io (2026-11-09): https://docs.cypress.io/app/guides/cross-browser-testing

    - Changelog, Cypress.io (2026-10-28): https://docs.cypress.io/app/references/changelog

    - Release notes, Microsoft (Playwright) (2026-11-09): https://playwright.dev/docs/release-notes

    - Browsers, Microsoft (Playwright) (2026-11-09): https://playwright.dev/docs/browsers

    - Supported languages, Microsoft (Playwright) (2026-11-09): https://playwright.dev/docs/languages

    - WebDriver BiDi, Selenium project (Software Freedom Conservancy) (2026-08-15): https://www.selenium.dev/documentation/webdriver/bidi/

    - WebDriver BiDi, W3C Browser Testing and Tools Working Group (2026-11-06): https://www.w3.org/TR/webdriver-bidi/

    - Selenium Grid, Selenium project (Software Freedom Conservancy) (2026-11-03): https://www.selenium.dev/documentation/grid/


    Follow the show: https://theplatformradar.com

    Episode page: https://theplatformradar.com/episodes/deep-dive-9-playwright-cypress-or-selenium-which

    Show More Show Less
    21 mins
  • PromQL mistakes that give you wrong p95s and blank error rates | The Platform Radar
    Nov 10 2026

    The p95 on your dashboard can be wrong, and the graph won't warn you. Learn how to tell whether the data or the query is at fault, before an incident depends on that number.


    You will learn what a histogram bucket really records, how rate and quantile get computed, and where those numbers go wrong. By the end, you can tell whether a wrong p95 or a strange error rate comes from the data or the query. You will also know what to change on Monday in your paging and dashboard queries so the number holds up during an incident.


    Chapters

    00:00 Intro

    01:20 Why the graph can't warn you

    03:40 What a bucket really records

    07:11 How rate and quantile get computed

    11:10 Where the numbers go wrong

    15:23 Queries to page and graph on

    19:02 What to change on Monday


    Sources (retrieved 2026-11-07)

    - Open standards in 2026: The backbone of modern observability, Grafana Labs (2026-03-18): https://grafana.com/blog/observability-survey-OSS-open-standards-2026/

    - Histograms and summaries, Prometheus Authors (2026-11-07): https://prometheus.io/docs/practices/histograms/

    - Query functions, Prometheus Authors (2026-11-07): https://prometheus.io/docs/prometheus/latest/querying/functions/

    - Recording rules, Prometheus Authors (2026-11-07): https://prometheus.io/docs/practices/rules/

    - Operators, Prometheus Authors (2026-11-07): https://prometheus.io/docs/prometheus/latest/querying/operators/

    - Instrumentation, Prometheus Authors (2026-11-07): https://prometheus.io/docs/practices/instrumentation/

    - Querying basics, Prometheus Authors (2026-11-07): https://prometheus.io/docs/prometheus/latest/querying/basics/

    - Prometheus 3.0 migration guide, Prometheus Authors (2026-11-07): https://prometheus.io/docs/prometheus/latest/migration/

    - Prometheus 3.16.0 release notes, Prometheus project (2026-11-04): https://github.com/prometheus/prometheus/releases/tag/v3.16.0

    - pint check: promql/rate, Cloudflare (2026-11-07): https://cloudflare.github.io/pint/checks/promql/rate.html


    Follow the show: https://theplatformradar.com

    Episode page: https://theplatformradar.com/episodes/deep-dive-8-the-promql-mistakes-behind-wrong

    Show More Show Less
    21 mins
  • .NET 8 patches stop Nov 10. Your deploys won't warn you | The Platform Radar
    Nov 3 2026

    On Lambda, dotnet8 has nine months between the day patches stop and the day deploys get blocked, and nothing in your pipeline flags that gap. Here is how to sort your .NET 8, Python 3.10 and Node 22 workloads by the date that actually matters.


    November 10 ends patches for .NET 8 and .NET 9, and your pipeline will keep shipping green builds anyway. This deep dive explains why you should plan around the vendor's end-of-support date, not the date your platform blocks deploys. It covers what running unpatched looks like in practice, and how to pick an upgrade target your platform actually runs in production for .NET 8, Python 3.10 and Node 22 workloads.


    Chapters

    00:00 Intro

    01:38 Why this week

    04:17 What November 10 means for .NET

    08:03 Why green deploys mislead you

    11:49 What unpatched looks like

    15:34 Pick what your platform runs

    19:03 Monday morning


    Sources (retrieved 2026-10-31)

    - .NET 8 and .NET 9 will reach End of Support on November 10, 2026, Microsoft .NET Blog (2026-06-29): https://devblogs.microsoft.com/dotnet/dotnet-8-9-end-of-support/

    - .NET and .NET Core Support Policy, Microsoft (2026-10-25): https://dotnet.microsoft.com/en-us/platform/support/policy/dotnet-core

    - Migrate C# apps from the in-process model to the isolated worker model, Microsoft Learn (2026-09-14): https://learn.microsoft.com/en-us/azure/azure-functions/migrate-dotnet-to-isolated-model

    - Lambda runtimes, Amazon Web Services (2026-10-31): https://docs.aws.amazon.com/lambda/latest/dg/lambda-runtimes.html

    - Introducing public preview runtimes on AWS Lambda, starting with Node.js 26 and Python 3.15, AWS Compute Blog (2026-08-15): https://aws.amazon.com/blogs/compute/introducing-public-preview-runtimes-on-aws-lambda-starting-with-node-js-26-and-python-3-15/

    - Cloud Run functions runtimes, Google Cloud (2026-10-31): https://docs.cloud.google.com/run/docs/runtimes/function-runtimes

    - Node.js 26.13.0 'Lithium' (LTS), Node.js Project (OpenJS Foundation) (2026-10-28): https://nodejs.org/en/blog/release/v26.13.0

    - Node.js release schedule, Node.js Release Working Group (2026-10-31): https://github.com/nodejs/Release

    - Wednesday, October 21, 2026 Security Releases, Node.js Project (2026-10-21): https://nodejs.org/en/blog/vulnerability/october-2026-security-releases

    - Status of Python versions, Python Software Foundation (2026-10-31): https://devguide.python.org/versions/


    Follow the show: https://theplatformradar.com

    Episode page: https://theplatformradar.com/episodes/deep-dive-5-patches-stop-nov-10-your

    Show More Show Less
    22 mins
  • .NET 8 patches stop Nov 10. Your deploys won't warn you | The Platform Radar
    Nov 3 2026

    On Lambda, dotnet8 has nine months between the day patches stop and the day deploys get blocked, and nothing in your pipeline flags that gap. Here is how to sort your .NET 8, Python 3.10 and Node 22 workloads by the date that actually matters.


    November 10 ends patches for .NET 8 and .NET 9, and your pipeline will keep shipping green builds anyway. This deep dive explains why you should plan around the vendor's end-of-support date, not the date your platform blocks deploys. It covers what running unpatched looks like in practice, and how to pick an upgrade target your platform actually runs in production for .NET 8, Python 3.10 and Node 22 workloads.


    Chapters

    00:00 Intro

    01:38 Why this week

    04:17 What November 10 means for .NET

    08:03 Why green deploys mislead you

    11:49 What unpatched looks like

    15:34 Pick what your platform runs

    19:03 Monday morning


    Sources (retrieved 2026-10-31)

    - .NET 8 and .NET 9 will reach End of Support on November 10, 2026, Microsoft .NET Blog (2026-06-29): https://devblogs.microsoft.com/dotnet/dotnet-8-9-end-of-support/

    - .NET and .NET Core Support Policy, Microsoft (2026-10-25): https://dotnet.microsoft.com/en-us/platform/support/policy/dotnet-core

    - Migrate C# apps from the in-process model to the isolated worker model, Microsoft Learn (2026-09-14): https://learn.microsoft.com/en-us/azure/azure-functions/migrate-dotnet-to-isolated-model

    - Lambda runtimes, Amazon Web Services (2026-10-31): https://docs.aws.amazon.com/lambda/latest/dg/lambda-runtimes.html

    - Introducing public preview runtimes on AWS Lambda, starting with Node.js 26 and Python 3.15, AWS Compute Blog (2026-08-15): https://aws.amazon.com/blogs/compute/introducing-public-preview-runtimes-on-aws-lambda-starting-with-node-js-26-and-python-3-15/

    - Cloud Run functions runtimes, Google Cloud (2026-10-31): https://docs.cloud.google.com/run/docs/runtimes/function-runtimes

    - Node.js 26.13.0 'Lithium' (LTS), Node.js Project (OpenJS Foundation) (2026-10-28): https://nodejs.org/en/blog/release/v26.13.0

    - Node.js release schedule, Node.js Release Working Group (2026-10-31): https://github.com/nodejs/Release

    - Wednesday, October 21, 2026 Security Releases, Node.js Project (2026-10-21): https://nodejs.org/en/blog/vulnerability/october-2026-security-releases

    - Status of Python versions, Python Software Foundation (2026-10-31): https://devguide.python.org/versions/


    Follow the show: https://theplatformradar.com

    Episode page: https://theplatformradar.com/episodes/deep-dive-5-patches-stop-nov-10-your

    Show More Show Less
    22 mins
  • Patch PHP RCE now, Python drops eMudhra CA, Postgres disk trap | The Platform Radar
    Oct 29 2026

    Three of this week's five Python releases ship as source only, and the PHP advisory is critical. Here is the order to patch in, plus the Postgres setting most teams never check.


    A patch-heavy week for platform teams. You get a ranked upgrade list: the critical PHP remote code execution fix, the Python releases that stop trusting compromised eMudhra intermediate keys and which ones ship as source only, the tarfile and cookie fixes, one Postgres setting to check for crash leftovers on disk, and the Node 22 and Terraform 1.16.7 point releases. Then, where survey data says agent lock-in really sits.


    Chapters

    00:00 Intro

    01:24 PHP critical RCE: update now

    03:17 Python stops trusting eMudhra keys

    06:40 The other fixes: tarfile and cookies

    09:55 Postgres crash leftovers on disk

    13:52 Node 22 'Jod' LTS point release

    15:23 Terraform 1.16.7 patch

    16:51 Models get swapped, harnesses stay

    20:28 Number of the week, what's next


    Sources (retrieved 2026-10-28)

    - Security advisory: Critical remote code execution vulnerabilities, update now, php-news (2026-10-21): https://www.php.net/index.php#2026-10-21-5

    - Python 3.15.2, 3.14.10, 3.13.18, 3.12.16 and 3.11.18 are out, python-insider (2026-10-26): https://blog.python.org/2026/10/python-3152-31410-31318-31216-31118/

    - Christophe Pettus: All Your GUCs in a Row: remove_temp_files_after_crash, planet-postgres (2026-10-25): https://thebuild.com/blog/all-your-gucs-in-a-row-remove_temp_files_after_crash/

    - hashicorp/terraform v1.16.7, terraform-releases (2026-10-22): https://github.com/hashicorp/terraform/releases/tag/v1.16.7

    - nodejs/node 2026-10-21, Version 22.23.4 'Jod' (LTS), @aduh95, node-releases-gh (2026-10-21): https://github.com/nodejs/node/releases/tag/v22.23.4

    - Survey Says: You'll swap the model, but everything you build around it is staying put, stackoverflow-blog (2026-10-21): https://stackoverflow.blog/2026/10/21/survey-says-you-ll-swap-the-model-but-everything-you-build-around-it-is-staying-put/


    Follow the show: https://theplatformradar.com

    Episode page: https://theplatformradar.com/episodes/recap-4-php-rce-python-ca-distrust

    Show More Show Less
    24 mins
  • Your model has an expiry date: who gets paged at 2am | The Platform Radar
    Oct 27 2026

    The same model snapshot can retire months apart on OpenAI and on Azure, and an SDK upgrade can change the API version you send without anyone touching that setting. This episode shows how to find every model and API version you depend on before the pager finds them for you.


    A deep dive for engineering leaders who ship on hosted language models. We walk through why a model ID is a dated dependency that can expire on a different day on each platform, how settings like auto-upgrade and inactivity rules can change what you run without a deploy, why an SDK major version can quietly change your API version, and how to decide when a vendor's suggested replacement is ready for your users.


    Chapters

    00:00 Intro

    01:40 Why this week, and the one claim

    04:27 Same model, different dates

    07:56 When the vendor swaps it for you

    11:31 What breaks, and who gets paged

    14:58 The SDK picks your API version for you

    18:08 What to do on Monday


    Sources (retrieved 2026-10-24)

    - Deprecations, OpenAI (2026-10-14): https://developers.openai.com/api/docs/deprecations

    - Model deprecations, Anthropic (2026-09-30): https://platform.claude.com/docs/en/about-claude/model-deprecations

    - Versions, Anthropic (2026-10-21): https://platform.claude.com/docs/en/api/versioning

    - Claude Developer Platform release notes, Anthropic (2026-10-21): https://platform.claude.com/docs/en/release-notes/overview

    - Migrating to API version 2026-10-21, Anthropic (2026-10-21): https://platform.claude.com/docs/en/api/migration-2026-10-21

    - Foundry Models lifecycle and support policy, Microsoft (2026-07-24): https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirements

    - Model retirement schedule, Microsoft (2026-09-21): https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule

    - Model lifecycle, Amazon Web Services (2026-09-07): https://docs.aws.amazon.com/bedrock/latest/userguide/model-lifecycle.html

    - Gemini deprecations, Google (2026-10-22): https://ai.google.dev/gemini-api/docs/deprecations


    Follow the show: https://theplatformradar.com

    Episode page: https://theplatformradar.com/episodes/deep-dive-3-your-model-has-an-expiry

    Show More Show Less
    21 mins
  • PHP RCE, 8 bad certs, Copilot cutoffs: what to act on | The Platform Radar
    Oct 22 2026

    PHP and Pillow both carry remote code execution warnings this week, and eight compromised certificates turn one vendor's problem into a shared trust problem. The ranked checklist separates what to patch today from what can wait.


    A ranked checklist for platform teams. You will learn which runtime and library patches carry remote code execution warnings, why eight compromised eMudhra certificates are a shared trust problem, what to check in your trust stores and pinning, which Copilot model pins stop working and what replaces them, and where the Cisco advisory, Azure Monitor advanced metrics and the Grafana Knowledge Graph belong on your list.


    Chapters

    00:00 Intro

    01:18 PHP 8.5.12 and a critical RCE warning

    05:04 Eight compromised eMudhra certificates

    07:54 Pillow and Ghostscript injection

    11:00 Copilot drops six models

    14:04 Cisco advisory CVE-2026-20043

    16:19 Azure Monitor advanced metrics GA

    18:53 Grafana expands its Knowledge Graph

    21:44 Number of the week and what's next


    Sources (retrieved 2026-10-21)

    - GitHub Copilot models deprecated on October 19, 2026, github-changelog (2026-10-19): https://github.blog/changelog/2026-10-19-github-copilot-models-deprecated-on-october-19-2026/

    - [Launched] Generally Available: Advanced platform metrics for Azure Monitor, azure-updates (2026-10-20): https://azure.microsoft.com/en-us/updates?id=575329

    - PHP 8.5.12 Released!, radar (2026-10-21): https://www.php.net/archive/2026.php#2026-10-21-4

    - Full-stack observability in Grafana Cloud: More connections, richer context, and faster investigations for users and agents, radar (2026-10-21): https://grafana.com/blog/full-stack-observability-with-knowledge-graph/

    - 8 eMudhra certificates are compromised. Here's what you need to know, radar (2026-10-21): https://blog.cloudflare.com/emudhra-certificate-compromise/

    - CVE-2026-112900: Pillow: Argument Injection in Ghostscript call, potentially leading into RCE, radar (2026-10-21): https://github.com/advisories/GHSA-w889-685w-c8fh

    - CVE-2026-20043: As part of Cisco's ongoing commitment to proactive security and product quality, the Cisco owned..., radar (2026-10-21): https://github.com/advisories/GHSA-jvf5-h9mp-fw5q


    Follow the show: https://theplatformradar.com

    Episode page: https://theplatformradar.com/episodes/recap-2-php-rce-8-bad-certs


    The host's voice is an AI clone of Mat Caniglia's own voice, used with his permission.

    Hosted by Mat Caniglia, in an AI clone of his own voice.

    Show More Show Less
    25 mins
  • Coding agents can wipe prod: what actually stops them | The Platform Radar
    Oct 21 2026

    Your permission rules and hooks check the command text. They never check what the command will actually delete. This episode shows which safeguards hold when an agent runs a recursive delete, and which ones only look like they do.


    Two real deletions followed the same pattern, and this episode walks through it. You learn why permission rules and pattern-matching hooks are weaker than they look, and how to sort every safeguard around your coding agent into advisory or control. Then we cover sandboxing, how much access an agent's credentials should have, and what is left when every other layer fails. You finish with a checklist to run against your own setup this week.


    Chapters

    00:00 Intro

    01:35 Two deletions, one pattern

    04:32 Rules and hooks read text

    07:54 The sandbox the OS enforces

    11:24 Credentials sized to the task

    14:44 When everything else fails

    18:06 What to do Monday morning


    Sources (retrieved 2026-10-21)

    - Claude Code deleted ~600GB: unprompted `rm -rf` on a substitution that resolved to a drive root, GitHub / anthropics/claude-code (report by the affected user) (2026-09-18): https://github.com/anthropics/claude-code/issues/95426

    - Configure the sandboxed Bash tool, Anthropic (Claude Code docs) (2026-10-21): https://code.claude.com/docs/en/sandboxing

    - Configure permissions, Anthropic (Claude Code docs) (2026-10-21): https://code.claude.com/docs/en/permissions

    - Hooks reference, Anthropic (Claude Code docs) (2026-10-21): https://code.claude.com/docs/en/hooks

    - Careful Adoption of Agentic AI Services, CISA with ASD's ACSC and international partners (2026-05-01): https://www.cisa.gov/resources-tools/resources/careful-adoption-agentic-si-services

    - OWASP Top 10 for Agentic Applications for 2026, OWASP GenAI Security Project (2025-12-09): https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/

    - AI Incident Database, Incident 1469: PocketOS production database reportedly deleted by Cursor AI agent running Claude Opus 4.6, Responsible AI Collaborative (AI Incident Database) (2026-04-24): https://incidentdatabase.ai/entities/pocketos/


    Follow the show: https://theplatformradar.com

    Episode page: https://theplatformradar.com/episodes/deep-dive-1-what-actually-stops-a-coding


    The host's voice is an AI clone of Mat Caniglia's own voice, used with his permission.

    Hosted by Mat Caniglia, in an AI clone of his own voice.

    Show More Show Less
    21 mins