Repository navigation
Prepare v2.0.1 official release - #20
Merged
Merged
Conversation
…analysis + time-series-analysis examples
The packaging script now also builds artifacts/bestfit-frequency-claude-plugin.zip: a skills-only Claude plugin whose root, one folder down, holds .claude-plugin/plugin.json and the unchanged skill. The same folder lists itself in a one-plugin marketplace, so Claude Desktop can install it by upload and Claude Code from a local marketplace; claude plugin validate --strict passes on both manifests. The OpenAI plugin keeps its .codex-plugin layout, which the current Codex loader still reads. Both plugins move to 0.3.1 because the skill runner changed after the 0.3.0 packages were built. The install guide covers the new routes, and the recorded package hashes match the rebuilt archives.
Add entries for the mixture row-identity and complete-state undo fix, the point-estimator refresh of comparisons, time-series properties and the nonstationary chronology, and the invalidation of time-series-derived input data and threshold diagnostics when the source changes. Retitle the notes for the 2.0.1 release and drop the superseded mixture grid re-bind wording.
Changing the credible-interval width, or moving the time index past the record, of an estimated nonstationary univariate analysis queues the frequency reprocess and then the chronology rebuild. Since 7b3297e the frequency reprocess ends by refreshing the chronology's point estimate and raising its notification, so the view redrew the chronology while the rebuild had already published results without their mean or intervals, and the redraw's NullReferenceException closed the application. CreateChronologyResultsAsync now builds the results privately and publishes them complete. The point-estimate refresh updates the chronology only while the curve's extent still matches its intervals, leaving a moved time index to the chronology rebuild. The chronology plot draws only complete, aligned results.
Since 556d6b5 the grid holds MixtureDistributionRow references, which cannot be compared, so choosing Sort Ascending or Descending from the column-header menu threw inside the menu click and closed the application. Sorting also reordered the displayed rows away from the component order. The grid now hides that menu, as the other order-sensitive App grids do.
An input switched from block or peaks-over-threshold extraction to a USGS download keeps its saved series reference. Since b445ddf any edit to that series cleared the downloaded peaks, low-outlier settings, and processed state without an undo step. Only block and peaks-over-threshold inputs derive their observations from the series, so only they are cleared; the 2.0.0 clear of USGS data on a series replacement is gone as well.
Release-note lines for the three review fixes: the chronology is published complete and drawn only when aligned, the mixture grid has no header sort menu, and manual and USGS inputs keep their observations when a formerly linked series changes.
Preserve borrowed connection ownership and successful close boundaries while releasing owned resources after failure. Add element, collection, and project regression coverage plus saved-project persistence comparison tools.
Reconcile source choices without resetting bound selectors, restore collection subscriptions after docking reloads, and release overlay wrappers on final element detachment. Add 21 isolated WPF lifecycle scenarios covering retained fits, overlays, covariates, deletion, source edits, and control reuse. Final Debug build and XML gate passed; all 5,731 fast tests passed on their final runs. An initial Bivariate point-estimator test failure passed unchanged in isolation and on the full App rerun; its intermittent cause remains unconfirmed.
Reconcile PR #12 at f8df484 with documentation-verification-updates at ed6f39a, preserving both histories and Sadie Niblett's citation credit. Keep every current saved project, backup, dataset and scientific source artifact unchanged. Retain the current scientific explanations, tables, retirements and additional Python figure coverage. Integrate the desktop walkthroughs and 122 current native captures with per-view provenance, explicit dispositions for all 135 contributor PNGs, and a 40-path data disposition. Correct view selection, diagnostic labels, fractional/calendar ticks and citation quoting. Add reproducible native panel capture and source/receipt verification without production changes. Validation: 39 rendered pages with zero broken local links; complete protected/staged-tree hash audit; all captures visually reviewed; 5,790 required fast .NET tests and 29 focused Python checks passed; independent scientific, preservation and tooling review completed. No saved estimator reruns; existing POT diagnostic display fits disclosed.
8 tasks done
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Prepare RMC-BestFit and the RMC.BestFit model package for v2.0.1, integrating the release branch and plugin preparation changes.
Release Significance
Validation
Historical numerical verification receipts remain scoped to their recorded methods and revisions; the full numerical verification suite was not rerun. Package validation and synthetic workflow smoke checks do not establish scientific interval coverage, successful installation in a named client/account, or plugin-directory publication.
Release Follow-up
Merge this PR before publishing the draft v2.0.1 release, and tag the eventual main merge commit. Confirm the Windows/Linux plugin CI checks, attach the matching desktop ZIP and SHA-256 checksum, verify live updater behavior, and confirm NuGet/plugin asset workflows after publication. Plugin-directory submissions remain separate.