A talk at NDC Oslo 2026 walks through about 40 Claude Code features most people never use. I checked them against seven months of my own prompt history, ran the testable ones headless, and tried some interactive ones by hand. Ctrl+G and rewind stuck, /focus did not, and a permissions.deny rule on .env let a shell glob read the file anyway.
Wrapping a C handle in unique_ptr with a function-pointer deleter doubles the pointer to 16 bytes. Putting the function into the deleter type instead, with an alias template over decltype of a static lambda, brings it back to 8. Sizes checked on GCC 14.2, GCC 16.2, Clang 22.1 and MSVC 19.44, with live Godbolt sessions.
You built your GStreamer plugin with -fsanitize=address and gst-launch-1.0 says the element doesn’t exist. The runtime has to be in the process before anything else. One LD_PRELOAD fixes it, and one repeated -D flag quietly turns ASan off in colcon builds.
A variadic concat(args…) that builds its own “{}{}{}” pattern at compile time, written three ways: a plain loop, a ranges pipeline, and an index_sequence with a comma fold. fmt 12 and std::format both still check the generated pattern at compile time, but only if it is static constexpr. Verified on GCC 14.2, GCC 16.2, Clang 22.1 and MSVC 19.44.
A buffer’s PTS is running time, not wall time. Add the pipeline’s base time and you get the clock time the frame was captured at. With the monotonic system clock that is a std::chrono::steady_clock time point you can compare to IMU or telemetry timestamps directly.
The latency tracer measures source to sink. To measure source to some point in the middle, hang a fakesink off a tee. Without queues on both branches the slow branch leaks into your number, or hangs the pipeline. Measured on GStreamer 1.24 in Docker.
A Google Drive folder that shows up as a normal directory every time I log in, using rclone mount, FUSE and a systemd user service. No root, no fstab, and a unit file that passes systemd-analyze verify.
Two days of rebuilding my shell setup on Ubuntu — the tools, the zshrc, the gotchas (fdfind/batcat names, starship install path, losing a zshrc to a bad paste) and the drills I used to make it stick.
I benchmarked same-host ROS 2 Jazzy pub/sub latency for 1 KB, 1 MB and 4 MB messages across six transport setups. Cyclone DDS + iceoryx with loaned messages moved 4 MB at an 83 µs median. Getting there took two settings nobody turns on by default, and the Fast DDS defaults fall over above 512 KB.
ZII designs every type so all-zero bytes is a valid default, and pairs it with arenas that free a whole group at once. RAII ties cleanup to each object. Here is how the two differ, where ZII falls short, a 20-line arena, and a nanobench run of 1000 allocations per frame: arena, std::pmr, malloc, calloc and new, with a live Godbolt session. Plus a struct-of-arrays layout, a Cortex-A78AE run, and the Jetson Orin GPU, then one CPU+GPU frame that uses all of it: 3.3x faster per frame.