jaeger 1.7.0

This is marked as a minor version although it should not have any visible changes, but the codebase has significantly changed and there's potential for regression issues that are better tracked as a clearly different version.

⚙️ Engineering

• #202 This started as a quick rewrite of some parts to use polars and ended as a mid-to-large refactor of significant parts of the code, especially the Configuration and Assignment classes.

  The main highlights are:

  • Dropped support for Python 3.9 and extended support up to 3.12. For Python >=3.11 the 1.4.0b1 version of kaiju is used.
  • The code does not use pandas anymore, and polars data frames are used everywhere. jaeger still handles pandas dataframes when they are returned by other libraries (mostly from the FVCTransform code in coordio).
  • The Configuration and Assignment classes have been completely rewritten. Coordinate transformations code is now mostly in jaeger.target.coordinates. The new code should be significantly cleaner and easier to maintain.
  • AssignmentData* has been renamed to Assignment*.
  • Some modest efficiency improvements to the coordinate transformations in Assignment. Before some conversions from ICRS to wok and vice-versa were done on a per-target bases. Now they are doing for all the targets at once, but the bottleneck is still the conversion between wok and positioner (and vice-versa) which has to be done as a loop for each target.
  • Simplified the singleton patter for FPS.
  • Significantly extended the test suite. Now Design/Configuration/Assignment and FVC are reasonably covered.
  • Added a test database for CI testing.
  • Added a configuration_to_dataframe function that generates a confSummary-like dataframe that could be saved to sdsscore as Parquet (currently not doing that).
  • Added ra/dec/alt/az_observed to confSummary.
  • Moved all codebase from python/ to src/.

jaeger 1.6.4

🔧 Fixed

• Change LCO expected scale factors after IMB modifications: apply only to LCO.

jaeger 1.6.3

✨ Improved

• Add --extra-epoch-delay to jaeger configuration load and preload. This parameter adds an extra delay to the configuration epoch. It is mainly used by HAL when it preloads a design ahead of time

jaeger 1.6.2

🏷️ Changed

• Change LCO expected scale factors after IMB modifications.

jaeger 1.6.1

✨ Improved

• Reset cherno offsets when a design gets loaded.
• Use coordio 1.9.2 with sdss-sep.

🔧 Fixed

• Fix some warnings and retries during broadcasts when a robot has been marked offline.
• Fix future deprecation in Pandas by downcasting columns in the assignment fibre data.
• Fix docs builds.

jaeger 1.6.0

🔥 Breaking changes

• Deprecate Python 3.8.

🏷️ Changed

• Use sdsstools version of yanny.

🔧 Fixed

• Ensure that `fvc_image_path`` is populated.

jaeger 1.5.0

🚀 New

• #200 Save copy of confSummary files to $SDSSCORE_TEST_DIR, if present. For testing purposes only, for now.
• Modify default ZB orders passed to the FVC transformation and add flag --polids to jaeger fvc loop to manually set the orders.
• Increased timeout for jaeger configuration preload.

✨ Improved

• #199 Subtract FVC dark frame during FVC image processing.
• #200 Initial test to update files in sdsscore_test.
• #201 Move FVC dark frames to calibration folder where they won't be deleted in the future.
• Updated call to coordio's object_offset().
• Chiller: do not depend on IEB when setting absolute values.

🔧 Fixed

• If measured alpha vs reported alpha are on either side of the wrap at 360 deg, then adjust the offset for the FVC loop.
• Update IEB info before calling write_proc_image().

⚙️ Engineering

• Lint using ruff.

jaeger 1.4.0

🚀 New

• Allow to specify the wavelength to use for BOSS and APOGEE fibres when creating a configuration. Using jaeger configuration load --boss-wavelength or --apogee-wavelength will specify the wavelength to use for atmospheric refraction and affect the positioning of the fibre. The closest valid focal plane model is used in this case. Only assigned fibre of the specified fibre type are affected.

🏷️ Changed

• Changed the default value of the safety factor for the offset function to 1.

jaeger 1.3.4

🚀 New

• Add flag --offset-min-skybrightness to jaeger configuration load/preload to set the coordio.utils.offset_definition offset_min_skybrightness parameters. Defaults to 0.5.

🏷️ Changed

• Add a 3 minute timeout for jaeger configuration preload (normal configuration load does not time out).
• COS-103 Add locked_alpha and locked_beta keywords when a collision happens.

jaeger 1.3.3

✨ Improved

• Add a delay after STOP_TRAJECTORY or SEND_TRAJECTORY_ABORT. Those commands are issued with timeout=0 so they complete immediately. It seems that if one sends another command immediately after them the new command times out. It's unclear if that happens because the replies from both commands clog the CAN buffer or because of some issue at the firmware level. Adding a 0.5 second delay to allow positioners to reply seems to fix the issue.