What’s New#

A short summary of the user-visible changes since the previous APS run. This page is updated once per APS run cycle (three runs per year); older entries are kept for reference.

For the full commit history see GitHub.

Run 2026-2 (Spring/Summer 2026)#

Covers changes merged since PR #30 (“Update repository to match latest apsbits”), which was the baseline for the 2026-1 run. Roughly 15 PRs landed during the shutdown — this section groups them by user-visible theme.

Sections are ordered by how often a typical user is likely to notice the change — restart workflow first, day-to-day scan helpers next, then per-experiment setup, then the lower-level work that’s mostly transparent.

Restarting a session — one line instead of a startup file#

The setup helpers (pr_setup, energy.tracking_setup, counters.plotselect, undulator_setup, qxscan_setup.load_params_json) now auto-save their values into RE.md["session_state"] every time you call them. After a bluesky restart inside the same experiment, a single import re-applies everything:

import id4_common.macros.startup_common  # noqa: F401

That runs experiment_resume() followed by restore_session_state() and prints a per-knob status. The hand-rolled startup file is now the exception, not the rule. See Restarting a Session.

A new curated star-import surface lives at id4_common.macros.macros_api — one import line for every public scan plan, peak-finder, and session singleton, kept stable across internal package reorgs. Three prebuilt motor-shortcut modules (shortcuts_4idg_euler, shortcuts_4idg_hp, shortcuts_4idh_9T) bind diffractometer / magnet sub-devices into __main__ under short names.

PRs: #65.

New peak-finding plans#

cen / com / maxi / mini are now plans you can call inside macros — they read the most recent scan from the catalog, compute the requested statistic (FWHM midpoint / centroid / max / min), and move the positioner there:

RE(cen())                                # FWHM midpoint of last scan
RE(com(positioner=tabx))                 # project a 2-D scan onto its axis
RE(maxi(scan_id=42, detector="hI2-apd"))

Backed by apstools.utils.xy_statistics + scipy.signal / scipy.ndimage. Works for both 1-D scans and 2-D grid_scan / rel_grid_scan (with noise-robust axis derivation that handles encoder jitter). Diagnostic-only versions: peak_pos, pmax, pmin.

The earlier apstools.lineup2-backed implementations are still available as cen2 / maxi2 / mini2 from id4_common.plans.peak_position_legacy.

See Peak-Finding Plans.

PRs: #62.

Temperature controllers — tc / ts / te#

A new temperature_setup(label) helper picks the active controller from a labelled lookup table and injects three names into the session:

temperature_setup("g")        # 4IDG LakeShore 336
temperature_setup("g-340")    # 4IDG LakeShore 340
temperature_setup("h-9T")     # 4IDH 9 T magnet VTI
mv tc 295                     # write the setpoint
te(295)                       # equivalent shortcut
ts.get()                      # read the sample temperature

ts is added to the baseline so the sample temperature lands in every scan automatically. Adding a new controller is a one-line edit to TEMPERATURE_CONTROLLERS in id4_common.utils.temperature_setup.

PRs: #57, #61.

CRL focusing — new shortcuts and microns handle#

  • crl_setup(hutch) switches the CRL sample-position offset between 4IDG and 4IDH ("g" / "h" / interactive).

  • crl_size(focal_size) sets the focal size in µm (with < 5 µm triggering crl.minimize_button instead).

  • mov crl.beamsize 10 writes 10 µm directly to the underlying focalSize PV. The previous EpicsSignal with write_pv would hang waiting for fSize_actual to converge; the new BeamsizeSignal Device returns immediately. The raw meter-valued PVs are still reachable as crl.beamsize.setpoint / crl.beamsize.readback.

The internal device class was renamed transfocator → crl across the codebase. The device is now crl in the registry; old transfocator references should be updated.

PRs: #58, #61.

Reciprocal-space scans (4IDG)#

Five new plans for the diffractometer (huber_euler / huber_hp):

RE(hscan(start, stop, npts, time))
RE(kscan(start, stop, npts, time))
RE(lscan(start, stop, npts, time))
RE(hklscan(h0, h1, k0, k1, l0, l1, npts, time))
RE(psiscan(psi_start, psi_stop, npts, time))   # rotate around Bragg vector

All accept dichro=True / lockin=True. Together with theta0, constraint cut points, and setor0 / setor1 / or_swap / Sync_UB_Matrix updates, the 4IDG session migrated fully from hklpy to hklpy2. The legacy hklpy code path was removed.

Two new devices were added to the 4IDG list: ringlight (sample camera ring light) and piezo_jena (Piezo-Jena nanopositioner stage).

PRs: #41, #45, #54.

Experiment lifecycle#

experiment_setup was refactored: it now auto-detects whether DM is reachable and falls back to dserv automatically with one warning (no need to remember skip_DM=True); fixed the reset_scan_id behaviour; added YAML persistence for restart; first unit-test suite in the repo.

experiment_resume() and experiment_load_from_scan(scan_id=-1) (renamed from the earlier experiment_load_from_bluesky) are the two ways to pick up where you left off — one from the YAML snapshot, one from a previous Bluesky run document.

PRs: #52.

Devices — baseline cleanup + missing devices wired up#

Several rarely-changed signals were demoted from kind="hinted" / kind="normal" to kind="config" so the baseline stream is no longer flooded with values that don’t change over an experiment.

Devices that existed in code but weren’t in devices.yml were added: srs810 (SRS-810 lock-in), gslt / hslt (4IDG/4IDH JJ slits), ringlight, piezo_jena, gmag (Kepco PS), comp / decomp (HP controllers), laser (Ventus laser).

PRs: #64.

Bug fixes and infrastructure#

  • SPEC writer now produces files that pymca can parse for qxscan and other array-metadata scans (no more bare line wraps inside #MD blocks; #S lines collapse arrays to <array len=N>).

  • load_device(name) is now the single canonical entry point for both first-load and reconnect (used to skip already-registered devices with a warning).

  • HDF1 priming wired up for Lambda250kDetector, VortexXspress34, VortexXspress37, VortexDante1, VortexDante4 — the first scan after a fresh IOC start no longer fails at staging.

  • Bluesky session logs now go to the central log dir specified by LOGGING.LOG_PATH in iconfig.yml (with automatic fallback to <cwd>/.logs/ when the central path is unwritable).

  • load_vortex has parity with load_device: always assigns the device to __main__, runs _post_connect_setup, primes HDF1.

  • Energy tracking prompt accepts 0 to disable tracking on every device.

  • pr_setup memory leak fixed; fixQ reset on scan abort fixed; scan hints fixed; predict_save_path and use_scalers fixes.

PRs: #37, #53.

Plans module split (mostly transparent)#

id4_common.plans.local_scans was split into focused modules so imports inside macros are tidier:

New module

Plans

base_scans

lup, ascan, count, qxscan, th2th

grid_scans

grid_scan, rel_grid_scan

hkl_scans

hklscan, hscan, kscan, lscan, psiscan

move_plans

mv, mvr, abs_set (with magnet911 staging)

peak_position

cen, com, maxi, mini, peak, peak_pos, pmax, pmin

peak_position_legacy

cen2, maxi2, mini2

id4_common.plans.local_scans still exists as a back-compat shim that re-exports from each of the above, so existing macros keep working without changes. New macros should prefer from id4_common.macros.macros_api import *, which is the curated surface and is kept stable across internal reorgs.

PRs: #56, #65.

Documentation#

The full documentation site was audited and refreshed:

  • New Useful Functions reference page covering every session-level helper (pr_setup, qxscan_setup, temperature_setup, restore_session_state, device loaders, …).

  • New “What’s New” page (this one), updated each APS run.

  • Several broken code snippets fixed (experiment_load_from_bluesky → experiment_load_from_scan, LakeShore .input.A.temperature / .output.one.setpoint → loop1.readback / loop1.setpoint, eight wrong device-class names in devices_guide.md).

  • peaks.cen vs peaks.com documentation fixed (they are not the same — peaks.cen is the FWHM midpoint, peaks.com is the moment-based centroid).

  • devices_reference.md filled in nine missing devices.

  • The autogenerated API reference is no longer tracked in git (regenerated on every build, was producing ~140 noisy diffs after every local sphinx-build).

PRs: #31, #66.