Resurrect previous NEWS.

NEWS.Rmd

@@ -429,3 +429,211 @@ Discrete Fourier Transform (FFT) verification on the data if wanted.
 ### `write_R2BIN()`
 * The function now supports the `verbose` argument, which can be used to
 suppress output to the terminal (#442, fixed in #444).
+
+
+# Changes in version 0.9.26 (2024-11-18)
+
+## Removed functions
+* Function `calc_Kars2008()` (deprecated since version 0.8.1) is now officially defunct,
+`calc_Huntley2006()` should be used instead (#252, fixed in #257).
+
+## Bugfixes
+
+### `analyse_pIRIRSequence()`
+* The function crashed with a object merge error if run in a loop because of a `merge_RLum()` error. This
+was caused by a regression while implementing the `n_N` calculation in `plot_GrowthCurve()`. Potentially
+affected was also `analyse_SAR.CWOSL()`.
+* The function now shows a warning and sets `plot = FALSE` when option
+`plot.single = FALSE` is set but the device size is too small. This should
+prevent "Figure margins too large" errors.
+* The function will not crash anymore during the plotting in another edge
+case related to single grain data.
+
+### `analyse_SAR.CWOSL()`
+* The function crashed for mixed set of `RLum.Analysis-class` objects
+typically observed for single grain data, e.g., one object contains only
+TL curves while all others contain the OSL/IRSL data from the grain. As the
+function filters those records automatically, it crashed because the aliquot
+number assignment internally relied on the length of the object (regression,
+#373, fixed)
+
+### `analyse_SAR.TL()`
+* The function now produces a more correct `rejection.criteria` data frame
+(#245, fixed in #246).
+* Several edge cases that led to crashes have been fixed (#147, fixed in #247).
+
+### `get_RLum()`
+* When the function was used on a list of `RLum.Analysis-class` objects with the argument `null.rm = TRUE` it would
+remove all `NULL` objects, but not elements that became `list()` (empty list) during the selection; fixed.
+* Fix an edge case that caused a rather non-expected, more visible output
+problem. When curves were selected via `recordType` on `RLum.Analysis-class`
+objects (or a list of them) containing only a single `RLum.Data-class`
+object, the function returned the `RLum.Data.Curve-class` object *regardless*
+of the selection in `recordType`. In other words: if a user tried
+`recordType = "TL"` on an `RLum.Analysis-class` object that contained only a
+single IRSL curve, the function would still return that single IRSL curve
+instead of an empty element. The reason for this behaviour was a poor attempt
+to deal with `NA` in the `recordType` name that led to missing values and
+unexpected behaviour for a logical comparison. Now, before the subset happens,
+`NA` values in `recordType` are converted to `"NA"` (a character), allowing us
+to fix the incorrect subsetting.
+
+### `plot_RLum.Data.Spectrum()`
+* Add support for `lphi` and `ltheta` light direction arguments for `plot.type = "persp"`.
+* Fix the reason for the unclear warning `In col.unique == col :  longer object length is not a multiple of shorter object length`
+
+### `write_R2BIN()`
+* Recently, non-ASCII characters in comments or file names have become more
+common and that led to crashes during the file export. To avoid this, now all
+non-ASCII characters are replaced by `_` before writing them to the BIN/BINX
+files.
+* The function now returns the file path of the export.
+* Fix a bug that left connections open if the function crashed.
+
+## Internals
+
+* Two new internal functions `.throw_warning()` and `.throw_error()` sometimes flushed the
+terminal with messages if called (internally) in particular circumstances. Now we maintain a
+stack of function names, so that at any time we can report correctly the name of the
+function where an error or a warning is thrown (#254, fixed in #256).
+
+
+# Changes in version 0.9.25 (2024-09-12)
+
+**This package version requires R >= 4.3**
+
+## New functions
+* `read_HeliosOSL2R()`: New import function to support the new zero rad Helios luminescence reader
+and its `.osl` file format. The output is an `RLum.Analysis-class` object.
+
+## Bugfixes
+
+### `analyse_baSAR()`
+* Fix #183 addressing an edge-case crash when the function was called on
+an object containing no records of the appropriate type (#184, @mcol).
+
+### `analyse_FadingMeasurements()`
+* Add `...` support to disable the blue trend line via `plot.trend = FALSE`
+* Fix #175 to deal gracefully with the case when the size of Lx and Tx
+doesn't match (#192, @mcol).
+
+### `analyse_portableOSL()`
+* Function did not respect argument `main`; fixed.
+
+### `convert_Wavelength2Energy()`
+* Fix #133 addressing cases where R throws an uncontained error when R drops
+the object structure in very rare cases (#134; thanks @mcol)
+
+### `calc_Huntley2006()`
+* Improve success rate of GOK fitting of the unfaded data using a two step approach: first
+an exponential fit is applied then this values are used as start parameters.
+
+### `calc_Statistics()`
+* The skewness and kurtosis depended on number of MC samples (#122); fixed with #123 (thanks to @mcol)
+
+### `calc_TLLxTxRatio()`
+* Function crashed for `Tx.data.background = NULL` (#129); fixed with #130 thanks to @mcol
+
+### `calc_WodaFuchs2008()`
+* The function now officially supports numeric vectors and single-column
+data frames as input (#200).
+* The function computed the number of breaks for the histogram incorrectly
+(#197, fixed with #198).
+* The function now supports more types of `RLum.Results` objects without
+crashing, although if the object contains only one data point it will stop
+immediately to avoid problems with `nls()` (#199, fixed in #219).
+
+### `extract_IrradiationTimes()`
+* If a BIN/BINX-file is provided, the function will now check that it contains the
+same amount of data as the corresponding XSYG file a bit earlier than before,
+thus avoiding a possible crash (#228, fixed in #229).
+
+### `fit_CWCurve()`
+* Argument `output.path` has been removed, and a warning is raised when
+attempting to use it (#207, fixed in #209).
+
+### `fit_EmissionSpectra()`
+* Parameter `input_scale` was not correctly propagated when the function would
+self-call (#160, @mcol).
+
+### `fit_OSLLifeTimes()`
+* The validation of the minimum dataset size didn't account for the use of
+the `signal.range` argument (#182, fixed by #195, @mcol).
+
+### `fit_SurfaceExposure()`
+* Fix #162 to remove a dimension mismatch if the input data contained `NA`s,
+which would generate unexpected warnings (#163, @mcol).
+* The function doesn't stop anymore with an error if called on an
+`RLum.Results` object (#165, @mcol).
+
+### `plot_AbanicoPlot()`
+* Argument `plot.ratio` will now throw an error on non-positive numerical
+values (#221, fixed in #222).
+* The function doesn't crash anymore when setting `interactive = TRUE` (#220,
+fixed in #233).
+
+### `plot_DetPlot()`
+* Argument `signal.integral.max` is now enforced to be greater than
+`signal.integral.min`, as otherwise the computation of the number of channels
+would produce `Inf` (#203, fixed in #206).
+* Fix a crash when using option `analyse_function = "analyse_pIRIRSequence"`
+(#210, fixed in #211).
+
+### `plot_GrowthCurve()`
+* The function now calculates the relative saturation (`n/N`) using the ratio
+of the two integrates. The value is part of the output table.
+* Argument `na.rm` has been removed: as of version 0.9.23, it was defunct
+and only accepted `TRUE` as valid value and produced an error otherwise,
+so there is no effective change in behaviour (#137, fixed in #214).
+
+### `plot_Histogram()`
+* The function doesn't crash anymore when setting `interactive = TRUE` (#186,
+fixed in #231).
+
+### `plot_KDE()`
+* It now officially supports numeric vectors and single-column data frames,
+for which it assumes that the De error at each measurement is 10^-9 (#189,
+fixed in #194, @mcol).
+
+### `plot_NRt()`
+* The function reports an helpful message rather than crashing when applied
+to an object of unexpected type or when there is a mismatch in time values
+(#177, fixed with #179 by @mcol).
+
+### `plot_RadialPlot()`
+* The function doesn't crash anymore when a single-column data frame is
+provided (#191, fixed in #212).
+
+### `plot_RLum.Data.Analysis()`
+* The function now supports all arguments from `plot_RLum.Data.Spectrum()`; before it
+had only basic functionality for `RLum.Data.Spectrum-class` data.
+
+### `plot_RLum.Data.Spectrum()`
+* The plot function can now handle non-increasing column values for plotting (with a warning).
+
+### `read_BIN2R()`
+* `ignore.RECTYPE` now supports numeric values, e.g., 128. Records for this type
+will be ignored during import.
+* BINX-files with `RECTYPE = 128` will not crash anymore, thanks for asking Anna-Maartje Boer and replying Karsten Bracht.
+* The function now stops graciously when attempting to read an empty file
+(#225, fixed in #226).
+
+### `read_PSL2R()`
+* The function is out of the beta status, hence the flag was removed.
+* The `RLum.Analysis-class` object returned by the function gained a new element `Sequence`, which is a data frame.
+with the measured sequence. This way, if the original sequence was lost, it can still be extracted from the `.psl` data.
+* If no `.psl` file was found the function got trapped in an infinite loop (#127); fixed with #128 (thanks to @mcol)
+
+### `read_XSYG2R()`
+* Fix spectrometer data import for basically broken files.
+
+### `read_Daybreak2R()`
+* Fix #135 to improve detection of non-ASCII files (#140, @mcol).
+
+### `Risoe.BINfileData2RLum.Analysis()`
+* Fix a crash when reading an empty Risoe.BINfileData input (#215, fixed in
+#224).
+
+## Internals
+* New internal function + tests added `.get_named_list_element()`. It just does what
+the names says.