LoginSign Up
← Back to all sparks
M

mantis

ANALYTICS
Velocity0.0

Multiple Time Series Scanner. Contribute to ropensci/mantis development by creating an account on GitHub.

github.com/ropensci/mantis ↗

Thirteen months of rOpenSci review turned a pre-release into a 1.0 with a stable API

time-seriesreportingropensciapi-stabilitydata-quality
◆Current state
mantis builds interactive time-series reports — heatmaps, multipanel plots, alert tables — over routinely collected data, and its release history is essentially an rOpenSci peer-review log. The package went from a January 2025 pre-release that warned of breaking changes ahead, through submission and CRAN acceptance, to a 1.0.0 in October 2025 that locked the API. Since then it has shipped only two small fixes, the most recent chasing a dplyr 1.2.0 deprecation.
◆Where it's heading
The breaking changes cluster entirely below 1.0.0 and stop there: `period` became `timepoint_unit`, `save_directory`/`save_filename` became `file`, `function_call` became `expression`. Post-1.0 the work is defensive — stricter POSIXt validation, timepoint limits that clamp to the data rather than inventing plot points, daylight-savings handling. The feed reads as a package that has finished defining itself and is now maintaining compatibility with the tidyverse underneath it.
◆Prediction
With the API frozen and review complete, expect the next release to be another upstream-compatibility fix rather than new report types. There is no signal in these entries about planned feature work.

◆Recent moves

  1. 6mo ago

    dplyr 1.2.0 compatibility fix

    Silences case_when() warnings introduced by dplyr v1.2.0. Pure upstream-compatibility maintenance, invisible to anyone reading a mantis report.

    View source ↗
  2. 7mo ago

    rOpenSci acceptance, plus an NA timepoint-limit fix

    Records formal acceptance into rOpenSci and fixes one edge case where an NA as the first value of `timepoint_limits` broke bespoke_rmd_output(). The acceptance is the milestone; the code change is a single guard.

    View source ↗
  3. 9mo ago

    1.0.0 locks the API and tightens timepoint handling

    The 1.0.0 that closes the review cycle: `alert_custom()` swaps `function_call` for `expression`, `item_order` accepts any number of items, and timepoint handling is tightened so date-only values in DST-varying timezones stop generating phantom points in plots. It is the last release to break anything — the API is fixed from here.

    View source ↗
  4. 1y ago

    Bespoke reports return to the CRAN build

    Restores bespoke reports, which the previous release had stripped out to get a lighter package past CRAN. This is the capability that lets users embed mantis widgets in their own Rmd rather than taking the canned report, so the CRAN build is only now feature-complete.

    View source ↗
  5. 1y ago

    First CRAN submission, standard reports only

    The first CRAN submission, deliberately narrowed to standard reports. It also reworks output paths — a single `file` parameter replaces `save_directory`/`save_filename`, `add_timestamp` handles suffixing, and intermediate files move to tempdir() instead of littering the working directory.

    View source ↗
  6. 1y ago

    rOpenSci submission; period renamed to timepoint_unit

    The rOpenSci submission marker, carrying one rename: `period` becomes `timepoint_unit` in inputspec(). Breaking, but on a package still explicitly pre-stable.

    View source ↗