Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Fetching contributors…

Cannot retrieve contributors at this time

98 lines (71 sloc) 3.724 kb
# tikzDevice TODO #
- Fill colors are still shown for polylines and lines; This does not
affect the output but is inefficient.
- Line endings are still shown for circles and squares; This does not
affect the output but is inefficient.
- Output to raw R objects.
- Support for raster images.
## FIXED ##
- Fontsize and ps parameters are not taken into account when calculating scaling factors.
Scaling factors are now calculated using:
(font size / base font size ) * cex
As opposed to the old way which was just the value of cex.
- Polylines are filled as white
Added a parameter tikzInfo->polyLine that is set to TRUE when StyleDef
is called, the fill color is only set if tikzInfo->polyLine == FALSE
- Shapes are plotted with a default color border of black when it should
be transparent
See the solution below.
- The CheckAndSetAlpha() function provides inconsistent results for
example test3 is completely transparent but test4 is not though.
Allowed for the option to set fill and draw transparency separately
with an extra parameter to CheckAndSetAlpha()
- Fill and draw colors are redefined before every shape is drawn, this
works as a first cut but is VERY inefficient for large numbers of points
which are not changing colors. possibly use \begin{scope}\end{scope} if
Introduced the parameters oldFillColor and oldDrawColor in the
tikzInfo structure which are checked against the current color and
only redefined if different
- when using plot(1,1), all calls to text result in proper output with one
exception: the coordinates of the text nodes are nan.
This error occurred because TikZ_Setup neglected to set the initial
size of the font by assigning a value to startps inside the device
structure. Additionally, plot characters were not produced because the
initial foreground and background colors were not set.
- when using plot(1,1), the bounding box of the plot is almost drawn
correctly. The line travels through the correct points, then returns to
(0,0) which results in the right hand side of the plot having a funky
This bug was a result of incorrect handling of C's #&@!&^ 0 array
indexing by TikZ_Polyline.
- When using plot(1,1), all text labels are placed incorrectly. This may
be a result of neglecting to provide a serious attempt at calculating
string width.
This was indeed a result of not making an attempt to calculate
the width of LaTeX strings. This has been fixed by a somewhat
elaborate system call to the LaTeX compiler. This would be an
elegent solution except for the fact that we have to call the
compiler 100s of times...
- Adding legends to plots results in incorrect sizing and positioning.
See lengend test for an example of this behavoir.
This was caused be not implementing a full metricInfo function.
Basically labels in legends seemd to hang below the symbols they
were plotted next to. This was caused by returning fixed ascents
and descents that were too large. Now LaTeX is called and the
character in question is set into a named TikZ node, say A. The
ascent is calculated by TikZ as:
A.north - A.base
The descent as:
A.base - A.south
And the width as:
A.east - A.west
Jump to Line
Something went wrong with that request. Please try again.