Skip to main content

2026

Every post published in 2026, newest first. Pick a month to narrow it down.

Posts
1,054
Topics
114
Reading time
128 h
Active months
6

Posts 121–150 of 1,054 · November 11–13, 2026

Page 5 of 36

November 2026

802 posts

Friday, November 13, 2026

19 of 34 posts

  1. How to Set the MLflow Tracking URI in Python

    Point your Python code at the right MLflow tracking server with mlflow.set_tracking_uri or the MLFLOW_TRACKING_URI environment variable: which one wins, how to check the current value, and what MLflow 3.18 does when you set neither (a hidden mlflow.db in your folder). All output is real.

  2. How to Open the MLflow UI and Find Your Runs

    A tour of the MLflow 3.18 UI for machine learning: switch from GenAI to Model training mode, close the Assistant panel, find your experiment, read the runs table, and open a run's overview, metrics charts and artifacts. Real screenshots.

  3. How to Turn Off MLflow Telemetry

    Open-source MLflow collects anonymous usage data by default since version 3.2.0. Turn it off for scripts, the tracking server and the UI with MLFLOW_DISABLE_TELEMETRY or DO_NOT_TRACK, on macOS, Linux and Windows, and prove it's off with MLflow's own check.

  4. How to Change the MLflow Server Port

    Run the MLflow tracking server on any port with --port, check which program already uses a port, and point your code at the new address. Includes why port 5000 fails on a Mac, and two MLflow servers side by side. Real output from MLflow 3.18.0.

  5. How to Start an MLflow Tracking Server on Your Computer

    Start your own MLflow tracking server with uv run mlflow server: what every startup line means, where it stores runs (sqlite:///mlflow.db and mlartifacts), how to check it's ready with /health and /version, and how to stop it safely. Real output from MLflow 3.18.0.

  6. How to Install MLflow in a Python Project with uv

    Install MLflow 3.18 into a Python project with uv add: what the 88 packages are for, how to check the version, the difference between mlflow, mlflow-skinny and mlflow-tracing, and the agent hint MLflow prints inside AI coding tools. Real output on Python 3.15.

  7. How to Load Settings from Environment Variables and a .env File with pydantic-settings

    Learn the pydantic-settings syntax, then read a Python app's settings from environment variables and a .env file: defaults, which source wins, a prefix for your app's names, an alias for a standard name such as DATABASE_URL, true-or-false values, and the errors you get for bad values. Real output from pydantic-settings 2.16.3.

  8. How to Check the Installed Version of a Python Package with uv

    Four ways to see which version of a Python package your uv project uses, each with its syntax first: uv pip show, uv tree, the tool's own --version, and importlib.metadata from Python. Real output for Rich 15.0.0, pytest 9.2.0 and Ruff 0.17.5 with uv 0.13.9.

  9. How to Run a Python Script with uv run

    Learn the uv run syntax, then run Python files, project commands and tools with the project's own packages. Shows why plain python3 says No module named 'rich', how arguments reach your script, and how uv run updates the environment first. Real output from uv 0.13.9.

  10. How to Add a Package to a Python Project with uv

    Learn the uv add syntax, then add a package to a Python project: what uv writes to pyproject.toml and uv.lock, how version ranges work, and how to keep development tools such as pytest and Ruff in their own group. Real output from uv 0.13.9.

  11. How to Create a Python Project with uv

    Learn the uv init syntax, then create a ready-to-use Python project: a src layout, pyproject.toml, a pinned Python version, a .gitignore and a Git repository. Plus what --python, --no-package, --lib and --bare change, with real output from uv 0.13.9.

  12. Fixed: "TypeError: got an unexpected keyword argument 'squared'" in scikit-learn mean_squared_error

    mean_squared_error(y_true, y_pred, squared=False), the RMSE line in many older tutorials, fails in scikit-learn 1.9.1 with TypeError: got an unexpected keyword argument 'squared'. The argument was deprecated in 1.4 and removed; root_mean_squared_error replaces it.

  13. Fixed: "[Errno 48] Address already in use" When You Start the MLflow Server

    mlflow server --port 8080 stopped at once with ERROR: [Errno 48] Address already in use and The server process exited with code 3. Another MLflow server was already listening on 127.0.0.1:8080. Find it with lsof, then use it, stop it, or start the new server on a free port.

  14. Fixed: "MlflowException: API request to … failed with exception … Max retries exceeded" Connection Refused in MLflow

    An MLflow training run sat silent, printed seven urllib3 retry warnings over four minutes, then failed with MlflowException: API request failed, Max retries exceeded, Connection refused. No tracking server was listening at the tracking URI. Start the server or fix the URI, and fail fast with MLFLOW_HTTP_REQUEST_MAX_RETRIES.

  15. Fixed: SADeprecationWarning "The noload loader strategy is deprecated" When You Run pytest with MLflow

    pytest passed every test but showed SADeprecationWarning: The noload loader strategy is deprecated, from MLflow 3.18.0's SQL tracking store with SQLAlchemy 2.1. MLflow fixed it after 3.18.0. One filterwarnings line in pyproject.toml hides exactly this warning until you upgrade.

  16. Fixed: "Exception ignored on threading shutdown" KeyboardInterrupt When You Stop the MLflow Server

    Stopping MLflow 3.18.0's tracking server with Ctrl+C printed Exception ignored on threading shutdown and a KeyboardInterrupt traceback from Python's threading.py. It comes from the server's background job runner, the server still stops cleanly, and MLFLOW_SERVER_ENABLE_JOB_EXECUTION=false removes it when you don't need server jobs.

  17. Fixed: "StarletteDeprecationWarning: starlette.middleware.wsgi is deprecated" When You Start the MLflow Server

    MLflow 3.18.0's tracking server printed StarletteDeprecationWarning: starlette.middleware.wsgi is deprecated four times at startup. It comes from MLflow's own code with Starlette 1.10.0, the server works, and one PYTHONWARNINGS filter hides exactly this warning until MLflow updates.

  18. Fixed: "Access to localhost was denied" HTTP ERROR 403 When You Open the MLflow UI on a Mac

    On a Mac, an MLflow server on port 5000 started fine, but http://localhost:5000 showed Access to localhost was denied, HTTP ERROR 403. The answer came from macOS's AirPlay Receiver, which also listens on port 5000. 127.0.0.1 reached MLflow, and port 8080 avoided the clash.

  19. Fixed: "Invalid comparator '>' not one of" When You Search MLflow Runs by a Parameter

    Searching MLflow runs with params.n_estimators > "100" failed with Invalid comparator '>' not one of, in the UI and in mlflow.search_runs. MLflow stores parameters as text, which only takes =, !=, LIKE and ILIKE. Match exact values, or compare the numbers in pandas.

Wednesday, November 11, 2026

11 of 27 posts

  1. How to Add a Stop Button to an AG-UI Chat

    Let people stop a long AG-UI answer: call abortRun on the HttpAgent, swap Send for Stop while a run is going, skip the client's own abort event, and pass the cancellation on in your server route — tested with @ag-ui/client 1.0.5 and a Microsoft Agent Framework agent.

  2. How to Customize the Request an AG-UI HttpAgent Sends

    By default, HttpAgent from @ag-ui/client 1.0.5 posts the whole conversation, tools and state on every run, and no parentRunId. Override its requestInit method to send only what your server accepts — with the real request bodies before and after.

  3. How to Filter AG-UI Events Before They Reach the Browser

    An AG-UI server streams more than a chat page needs: the model provider's raw data in rawEvent, token usage, encrypted reasoning. Keep an allowlist of event types and drop rawEvent and usage in your server route — with real events from a Microsoft Agent Framework agent before and after.

  4. How to Keep a Chat Scrolled to the Newest Message in React

    Keep a growing chat or streamed answer scrolled to its newest line in React with an empty element at the end, a ref and scrollIntoView in a useEffect — compared side by side with the same list without it, in a real browser.

  5. How to Send a Message with Enter and Add a New Line with Shift+Enter in React

    Make a React textarea send its form on Enter and start a new line on Shift+Enter, without breaking Japanese, Chinese or Korean input: an onKeyDown handler with isComposing and requestSubmit, tested in a real browser.

  6. How to Open Markdown Links in a New Tab with react-markdown

    Replace the link element of react-markdown with the components option so every Markdown link opens in a new tab with rel noreferrer — and leave out the node prop, which otherwise ends up in the HTML as node="[object Object]". Real DOM output from react-markdown 10.1.0.

  7. How to Render Markdown Safely in React with react-markdown

    Show Markdown from users or AI models as real headings, lists, tables and links with react-markdown and remark-gfm — without dangerouslySetInnerHTML. A real page shows HTML inside the Markdown appearing as text and a javascript: link losing its address.

  8. How to Support Light and Dark Mode with CSS Variables

    Define every color once as a CSS variable, give the variables dark values inside prefers-color-scheme, and add color-scheme so form fields follow too — a page that switches with the operating system, with real light and dark screenshots.

  9. How to Cancel an Upstream Request When the Browser Disconnects in Next.js

    When a Next.js route handler calls another service, pass request.signal to fetch so the call ends the moment the browser leaves. A side-by-side run shows the upstream stopping after one second with the signal and working five seconds for nobody without it.

  10. How to Stream Server-Sent Events from a Next.js Route Handler

    Return a ReadableStream from a Next.js route handler to send Server-Sent Events as they happen: the data line format, the headers that stop buffering, cleaning up when the client leaves, and an EventSource page that shows each event — with real timings from Next.js 16.4.6.

  11. How to Validate a JSON Request Body in a Next.js Route Handler with Zod

    Check the JSON body of a POST request in a Next.js route handler with a zod schema: safeParse, a 400 answer with a message per field, bodies that aren't JSON, and what happens to extra fields — with real curl output from Next.js 16.4.6 and zod 4.6.7.