stringr
stringr keeps trading convenient guesses for predictable errors.
A side-by-side editorial comparison of jwst and rlang — release velocity, themes, recent moves, and the top alternatives to consider.
JWST's calibration pipeline extended adaptive trace modelling across its spectrographs
Version 3.0.0, the DMS B13.0 operational build, is the substantive release in this window. It extends the adaptive_trace_model step to NIRSpec MOS, fixed-slit and BOTS modes plus MIRI LRS, adds multiprocessing that cut one NIRSpec IFU case by roughly a factor of seven, and introduces chromaticity correction for NIRSpec IFU data via a new reference file type. It also removes several internal-only step parameters as breaking changes. The four release candidates preceding it contain only dependency pins and changelog freezes.
rlang moved tidyeval off R's private internals and onto official C API.
rlang is at 1.3.0, which rewrote hash() to walk objects itself rather than lean on R's serialiser — fixing stability against bytecode and shrinkable vectors, at the cost of invalidating every existing hash value. The release before it closed a multi-year effort: rlang and tidyeval are now fully backed by official C APIs of R, work the notes credit to collaboration with R core.
Version 3.0.0, the DMS B13.0 operational build, is the substantive release in this window. It extends the adaptive_trace_model step to NIRSpec MOS, fixed-slit and BOTS modes plus MIRI LRS, adds multiprocessing that cut one NIRSpec IFU case by roughly a factor of seven, and introduces chromaticity correction for NIRSpec IFU data via a new reference file type. It also removes several internal-only step parameters as breaking changes. The four release candidates preceding it contain only dependency pins and changelog freezes.
Development is organised around periodic DMS operational builds rather than continuous delivery, with release candidates used purely to freeze dependencies. The direction inside the pipeline is toward per-mode calibration sophistication - trace modelling and chromaticity corrections that were previously unavailable or mode-limited - alongside a steady cleanup of parameters that only ever existed for internal plumbing.
Expect adaptive trace modelling to keep expanding across the remaining instrument modes, and the multiprocessing work applied there to spread to other slow steps. Further breaking removals of internal-use parameters are likely while the 3.x major version is open.
rlang is at 1.3.0, which rewrote hash() to walk objects itself rather than lean on R's serialiser — fixing stability against bytecode and shrinkable vectors, at the cost of invalidating every existing hash value. The release before it closed a multi-year effort: rlang and tidyeval are now fully backed by official C APIs of R, work the notes credit to collaboration with R core.
The through-line across this whole window is one migration. Release after release retires something that depended on private R internals — env_browse(), env_unlock(), ns_registry_env(), the SEXP iterator now behind a compile flag — and replaces it with sanctioned API. The hash() rewrite in 1.3.0 is the same instinct applied to the serialiser: own the behaviour rather than inherit it.
With the C API migration declared complete in 1.2.0, the next releases are likely to be ordinary maintenance and type-checking additions rather than further defunct markings.
Other DevOps products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either jwst or rlang.
stringr keeps trading convenient guesses for predictable errors.
pyjanitor is folding its verbs into pandas groupby objects, one release at a time.
purrr finished a decade of deprecations and picked up a parallel backend.
PyTables opened a path around HDF5's filter pipeline, then chased Python's runtime.
R's API framework grew its serializer catalogue, then went quiet on features.
Dask's scheduler spent the year deleting deprecated API, not adding surface.
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. jwst is currently shipping more aggressively (velocity 3.8 vs 0.0), with 1 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. jwst is currently shipping more aggressively (velocity 3.8 vs 0.0), with 1 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other DevOps products to evaluate alongside.
Top jwst alternatives in DevOps are ranked by recent ship velocity. Browse the "jwst alternatives" section above for the current picks, or visit /alternatives/jwst-pipeline for the full list with editorial commentary on each.
Top rlang alternatives in DevOps are ranked by recent ship velocity. Browse the "rlang alternatives" section above for the current picks, or visit /alternatives/rlang for the full list with editorial commentary on each.