NEWS.md
Fix bugs resulted from the new released version of ggplot2 3.4.0
ggplot2 3.4.0, the aesthetic attribute size is no longer recommended to set the line width for all geometric objects. Most errors and warnings are caused by this.ggplot2, GeomCol (geometric layer geom_col) is now inherited from GeomBar (its supper class). Therefore, in the transformation, a ggplot with geom_col() layer will be converted to an interactive l_hist plot, which should not be (note that, only geom_bar and geom_histogram could be turned an interactive l_hist plot).Important Update: to transform a loon l_facet object to a ggplot object, instead of relying on the package patchwork to combine plots, the function facet_grid() or facet_wrap() is used.
A patchwork object can be transformed to a loon l_compound object.
In the previous versions of ggplot2 (< 3.3.5), both none and FALSE work in guides(). In the latest version (e.g., >= 3.3.5), FALSE is deprecated. Therefore, to set guides() in the package loon.ggplot, we use none.
Fix bugs:
To transform a ggplot object to a loon plot, if the points shapes are between 21 to 25, the variable “fill” is used as the loon points color; if the variable “fill” is not found or set as NA, then the variable “colour” is used.
To transform a ggplot barplot to a loon l_hist widget, the binwidth is set as 1 by default.
Important Changes: when we turn a loon compound object to a ggplot object, package patchwork is used (not GGally). The benefit is that the size of each plot can be different.
Fix a bug: a error is encountered when we transform a swapped loon histogram to a ggplot object.
A loon zenplot can be turned to a ggplot object via the function loon.ggplot() (or loon2ggplot()).
Fix a bug: in the previous versions, to transform a ggplot object to a loon plot, showItemLabels and itemLabels are failed to be passed into the l_serialaxes widget.
Set ggmulti as “Imports” rather than the “Depends”, and replace coord_serialaxes to geom_serialaxes in order to avoid potential failures (in coord_serialaxes).
Executing the function print.l_ggplot(), a message will be given to show the loon plot path name.
Class name modification: in the earlier versions, to transform a ggplot object with facets to a loon plot, the class of the returned object is c(“l_ggplot”, “l_compound”, “loon”). However, for an l_ggplot() function, the class of the returned object is c(“lggplot”, “gg”, “ggplot”) which is confusing. In this version, the class of the former one is c(“l_facet_ggplot”, “l_facet”, “l_compound”, “loon”); and the class of latter one is c(“l_ggplot”, “gg”, “ggplot”).
For an l_facet_ggplot object (transformed from the ggplot with multiple facets), the returned object is changed.
In the earlier versions, the returned object is a list of three components:
plots: a list of loon widgets;
facet: a list of four logical components, FacetWrap, FacetGrid, byCOLS and byROWS
titles: a list of three components, title of the whole widget, column subtitles and row subtitles.
In this version, the returned object is just a list of loon widgets, like other l_compound objects (e.g., l_facet, l_pairs).
Two changes are made when we transform a ggplot histogram to a loon l_hist widget:
If the ggplot histogram shows “density”, the area of each category (grouped by color) is 1; however, in an l_hist widget, the whole area is 1 and the area of each category is proportional to the counts. Thus, after transformation, the y limits are identical but the display could be very different. To get a better display, the y limits are released (may result the different visual displays) and a message is given.
Suppose the transformed loon histograms are joined into a linking group, the colors of each bin could be different from the original ggplot object. Therefore, the state colorStackingOrder will be reset.
New interactive component
scaleTo: used to change the region (i.e., scale to selected points, active points, or a specific geometric layer) of the loon plot.
active: determine which geom layer could be interactive or which points could be activated.
From loon to ggplot, a logical argument asAes is given (default is TRUE). If TRUE, the color, size, other aesthetics will be taken as variables and set in the function aes(); else they are the general aesthetics attributes.
Fix a bug: the NA of some n dimensional states should be checked before passing through.
A warning is given once multiple active layers are not set.
Remove the tidyverse dependency
When a loon widget –> a ggplot object, several things are changed here:
Add new features (e.g., ‘Andrews curves’);
If z is provided in the mapping aesthetics of geom_point, an l_plot3D object would be created.
In loon, if points have non-primitive glyphs (viz., text glyph, pointrange glyph, image glyph, polygon glyph, and serialaxes glyph), in the transformation, functions geom_imageGlyph(), geom_polygonGlyph(), geom_pointrangeGlyph(), geom_serialaxesGlyph() and geom_textGlyph() are deprecated and not maintained any more. Instead, we use the package ggmulti who provides functions geom_image_glyph(), geom_polygon_glyph() and geom_serialaxes_glyph() to draw the non-primitive glyphs (for the layer geom_text and geom_pointrange, they would be turned into an interactive layer automatically. To turn them as static, one can set the argument activeGeomLayers as 0).
To transform an l_serialaxes object, the function ggSerialAxes() is deprecated and ggmulti::coord_serialaxes() is used.
Some ggplot2 extensions provide new geom layers (e.g., geom_textpath). To transform these layers in loon, one can customize the function loonLayer().