Thread: [Plplot-cvs] SF.net SVN: plplot:[11831] trunk/ChangeLog.release

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Revision: 11831
          http://plplot.svn.sourceforge.net/plplot/?rev=11831&view=rev
Author:   hbabcock
Date:     2011-08-01 13:12:31 +0000 (Mon, 01 Aug 2011)

Log Message:
-----------
Repurpose legacy ChangeLog file as the ChangeLog.release file. Update for the 5.9.8 release.

Added Paths:
-----------
    trunk/ChangeLog.release

Copied: trunk/ChangeLog.release (from rev 11830, trunk/ChangeLog)
===================================================================

--- trunk/ChangeLog.release	                        (rev 0)
+++ trunk/ChangeLog.release	2011-08-01 13:12:31 UTC (rev 11831)
@@ -0,0 +1,7927 @@
+------------------------------------------------------------------------
+r11830 | hbabcock | 2011-08-01 08:54:13 -0400 (Mon, 01 Aug 2011) | 1 line
+Changed paths:
+   M /trunk/cmake/modules/plplot_version.cmake
+
+Update date and version number for the 5.9.8 release. Increment SOVERSION for the libplplotd and libplplotcxxd libraries to reflect backwards incompatible API changes made in this version.
+------------------------------------------------------------------------
+r11829 | airwin | 2011-07-31 04:58:36 -0400 (Sun, 31 Jul 2011) | 14 lines
+Changed paths:
+   M /trunk/scripts/comprehensive_test.sh
+
+Implement do_clean_as_you_go option to clean up as you go.  However, it
+does not work well, yet, because the custom commands OUTPUT files
+just refer to the files from the first example rather than all 33 so
+those first example files are the only ones cleaned (so far).
+
+ToDo (some time in next release cycle) implement complete OUTPUT files
+for custom commands.
+
+Change launch of ctest job to execution of ctest.  This probably takes
+longer (at least until we make ctest parallel), but it does allow
+clean up afterwards since this change makes ctest end become
+completely predictable within the scripts.
+
+
+------------------------------------------------------------------------
+r11828 | airwin | 2011-07-31 03:30:27 -0400 (Sun, 31 Jul 2011) | 3 lines
+Changed paths:
+   M /trunk/examples/java/Makefile.examples.in
+   M /trunk/examples/ocaml/Makefile.examples.in
+
+Add example 33 to those which are built for the traditional install
+tree.
+
+------------------------------------------------------------------------
+r11827 | airwin | 2011-07-31 02:49:51 -0400 (Sun, 31 Jul 2011) | 3 lines
+Changed paths:
+   M /trunk/examples/d/Makefile.examples.in
+
+Include 33rd example in list of those that are built in the traditional
+install tree approach.
+
+------------------------------------------------------------------------
+r11826 | airwin | 2011-07-29 15:50:14 -0400 (Fri, 29 Jul 2011) | 4 lines
+Changed paths:
+   M /trunk/ABOUT
+
+More tweaking including spelling issues found by ispell, some grammar
+tweaking on the questions, and mentioning
+http://www.ohloh.net/p/plplot/contributors.
+
+------------------------------------------------------------------------
+r11825 | hbabcock | 2011-07-29 13:34:46 -0400 (Fri, 29 Jul 2011) | 1 line
+Changed paths:
+   M /trunk/ABOUT
+
+Add a description of the PLplot roadmap (or lack thereof). Tweak the formatting a bit.
+------------------------------------------------------------------------
+r11824 | airwin | 2011-07-27 00:26:10 -0400 (Wed, 27 Jul 2011) | 2 lines
+Changed paths:
+   M /trunk/ABOUT
+
+Set appropriate properties for this file.
+
+------------------------------------------------------------------------
+r11823 | airwin | 2011-07-27 00:06:47 -0400 (Wed, 27 Jul 2011) | 3 lines
+Changed paths:
+   M /trunk/ABOUT
+
+Add an extensive section on usage statistics, a minor section
+on our developers, and many tweaks.
+
+------------------------------------------------------------------------
+r11822 | hbabcock | 2011-07-26 20:30:08 -0400 (Tue, 26 Jul 2011) | 1 line
+Changed paths:
+   A /trunk/ABOUT
+
+Initial version of a ABOUT file that provides a brief overview of the PLplot project.
+------------------------------------------------------------------------
+r11821 | smekal | 2011-07-19 11:46:38 -0400 (Tue, 19 Jul 2011) | 1 line
+Changed paths:
+   M /trunk/drivers/wxwidgets_gc.cpp
+
+The text clipping didn't work for wxWidgets 2.9.2/Cocoa/Mac since the wxRegion constructor used wasn't implemented. Only for this wxWidgets configuration another wxRegion constructor is used. Apart from that there was no problem using the new wxWidgets version (which soon will become 3.0.0) with the PLplot wxWidgets driver.
+------------------------------------------------------------------------
+r11820 | airwin | 2011-07-18 01:34:57 -0400 (Mon, 18 Jul 2011) | 18 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+Improve internal documentation of calculate_limits so that it
+is more obvious what is being calculated.
+
+Distinguish arguments for calls to calculate_limits more carefully
+to keep track of everything necessary for calculating offsets
+and sizes of bounding boxes for undecorated colorbar, decorated
+colorbar, and labelled and decorated colorbar.
+
+Implement bounding box calculation in draw_label so that the overall
+bounding box (for labelled and decorated colorbar) will (finally) be
+correct.  This change has been tested for long labels for plcolorbar
+in example 33 where both x and y bounding box limits are increased as
+well as the standard short labels for plcolorbar in example 33.
+
+This commit completes my planned work on the bounding box and
+position calculations for plcolorbar.
+
+
+------------------------------------------------------------------------
+r11819 | smekal | 2011-07-17 15:38:33 -0400 (Sun, 17 Jul 2011) | 1 line
+Changed paths:
+   M /trunk/drivers/wxwidgets_gc.cpp
+
+Fixed a memory allocation bug, which showed up if one resizes the window.
+------------------------------------------------------------------------
+r11818 | airwin | 2011-07-17 13:34:03 -0400 (Sun, 17 Jul 2011) | 12 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+Make second call to calculate_limits with appropriate transformation
+of input and output results.  Since additional coordinate system
+tranformations are involved the results are numerically different, but 
+the ndiff app confirms these changes are only for
+one unit in the last PostScript place.  That numerical result helps
+to verify that the correct transformations are being used for
+the second calculate_limits call.
+
+ToDo:  Implement bounding box calculation in draw_label so that the
+overall position and bounding box will (finally) be correct.
+
+
+------------------------------------------------------------------------
+r11817 | airwin | 2011-07-17 12:30:21 -0400 (Sun, 17 Jul 2011) | 29 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+Code reorganization in preparation for calculating complete bounding box
+and position of the labelled and decorated colorbar.
+
+Systematically use the mm form of bounding box limits,
+plsc->boxbb_xmin, etc., rather than the confusing mixture of that form
+and one where the limits are expressed in normalized subpage
+coordinates.
+
+Implement the static function, calculate_limits, to isolate in one
+place all the limit changes due to a changed bounding box.  The first
+call of this function is associated with changing the bounding box and
+corresponding position calculation from the undecorated colorbar to
+the decorated (ticks, numerical tick labels, and exponents) colorbar.
+The second call of this function (not yet used) will be associated
+with changing the bounding box and corresponding position
+calculationfrom the decorated colorbar to the labelled and decorated
+colorbar.
+
+However, the calculate_limits code is not yet fully
+understood/documented (since it just collected pre-existing code which
+was not fully understood/documented into a static function) so the
+second call is not used yet.
+
+As expected, the result is identical to the previous commit, i.e.,
+labels are currently ignored in the bounding box calculation and in
+the calculation of the position of that bounding box in the overall
+plot.
+
+
+------------------------------------------------------------------------
+r11816 | airwin | 2011-07-14 12:37:58 -0400 (Thu, 14 Jul 2011) | 3 lines
+Changed paths:
+   M /trunk/examples/c/x33c.c
+
+Decrease size of overall viewport so there is room to display labelled
+and decorated colorbar.
+
+------------------------------------------------------------------------
+r11815 | airwin | 2011-07-14 12:36:01 -0400 (Thu, 14 Jul 2011) | 10 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+Draw overall label relative to the bounding box of the rest of the
+decorated (including tick marks, numerical labels for ticks,
+exponents, and end caps) colorbar.  This change greatly simplifies
+the draw_box arguments and logic, and makes a much nicer-looking
+result.
+
+ToDo: Calculate complete bounding box of label + rest of the decorated
+colorbar.
+
+
+------------------------------------------------------------------------
+r11814 | airwin | 2011-07-13 18:49:07 -0400 (Wed, 13 Jul 2011) | 8 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+Reorganize plcolorbar code so the overall colorbar label is drawn with
+the separate static function draw_label.  This reorganization gives
+identical example 33 results to before and prepares the way to
+reposition the label to be centred relative to the bounding box of the
+inner decorated box and to also do bounding box calculations for the
+whole object (label + inner bounding box).
+
+
+------------------------------------------------------------------------
+r11813 | airwin | 2011-07-13 18:02:56 -0400 (Wed, 13 Jul 2011) | 17 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+Fix accident waiting to happen for get_subpage_per_mm (a static
+function which is called directly or indirectly by pllegend and plcolorbar).
+
+The only reason this function worked correctly before was it was always
+called after
+
+plvpor( 0., 1., 0., 1. );
+plwind( 0., 1., 0., 1. );
+
+But the point is the get_subpage_per_mm doesn't need to know anything
+about the viewport, and a call to plgspa returns all data required to
+calculate the x and y ratios of normalized subpage coordinates to mm.
+So get_subpage_per_mm can be substantially simplified, and that
+simplification insures that routine will give correct results when
+plvpor and plwind have been called with other than the above values.
+
+
+------------------------------------------------------------------------
+r11812 | airwin | 2011-07-12 17:52:45 -0400 (Tue, 12 Jul 2011) | 8 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+Fix memory management issues with draw_box (static function called by
+plcolorbar).
+
+Implement draw_box documentation.
+
+Improve plcolorbar documentation and parameter sanity checking/defaults.
+
+
+------------------------------------------------------------------------
+r11811 | airwin | 2011-07-12 01:48:39 -0400 (Tue, 12 Jul 2011) | 10 lines
+Changed paths:
+   M /trunk/src/plbox.c
+
+Implement bounding-box functionality for horizontal and vertical axes
+of custom labels following what is done in plbox.  The bounding-box
+results for custom labels look the same (good) as for more ordinary
+labels done with plbox in example 33.
+
+ToDo: For plcolorbar still need to implement adjustment of overall
+colorbar label position relative to plbox bounding box and implement
+bounding-box functionality for (moved) overall colorbar label.
+
+
+------------------------------------------------------------------------
+r11810 | airwin | 2011-07-12 00:16:54 -0400 (Tue, 12 Jul 2011) | 9 lines
+Changed paths:
+   M /trunk/src/plbox.c
+
+Update bounding box to include vertical axis exponents for plbox
+bounding box.  Example 33 confirms this improvement is working
+properly.  This completes the bounding box calculation for plbox!
+
+ToDo: For plcolorbar still need to implement bounding-box
+functionality for horizontal and vertical axes of custom labels and
+bounding-box functionality for overall colorbar label.
+
+
+------------------------------------------------------------------------
+r11809 | airwin | 2011-07-11 21:47:09 -0400 (Mon, 11 Jul 2011) | 7 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+For some reason, plgradient gives unreliable results when the x and y
+world coordinates have vastly different magnitudes.  Work around this
+issue, by changing the world coordinates of the direction perpendicular
+to the plcolorbar from {0., 1.} to {0., max_abs} where max_abs is
+the maximum of the absolute values of the minimum and maximum of
+the world coordinate range being plotted by the colorbar.
+
+------------------------------------------------------------------------
+r11808 | airwin | 2011-07-11 18:11:54 -0400 (Mon, 11 Jul 2011) | 13 lines
+Changed paths:
+   M /trunk/src/plbox.c
+
+Update bounding box to include both vertical and horizontal components
+of numerical labelling of vertical axes.  Example 33 confirms this
+improvement is working properly.  This completes (aside from the
+vertical axis exponents) the bounding box calculation for vertical
+axes of plbox which has been verified using example 33 with and without
+the "v" option in axis_opts.
+
+ToDo: Still need to implement vertical axis exponents for plbox
+bounding box.  For plcolorbar also need to implement bounding-box
+functionality for horizontal and vertical axes of custom labels and
+bounding-box functionality for overall colorbar label.
+
+
+------------------------------------------------------------------------
+r11807 | airwin | 2011-07-11 18:04:24 -0400 (Mon, 11 Jul 2011) | 2 lines
+Changed paths:
+   M /trunk/src/pllegend.c
+
+Fix small error in line-style restoration.
+
+------------------------------------------------------------------------
+r11806 | airwin | 2011-07-11 18:03:15 -0400 (Mon, 11 Jul 2011) | 3 lines
+Changed paths:
+   M /trunk/examples/c/x33c.c
+
+Use negative numbers in range to make sure bounding-box calculation is
+still good for that.  (It was.)
+
+------------------------------------------------------------------------
+r11805 | airwin | 2011-07-11 17:33:12 -0400 (Mon, 11 Jul 2011) | 3 lines
+Changed paths:
+   M /trunk/examples/c/x33c.c
+
+Show how to test parallel axis labels for vertical axis of colorbars,
+but comment out (now that I have finished that test).
+
+------------------------------------------------------------------------
+r11804 | airwin | 2011-07-10 13:27:00 -0400 (Sun, 10 Jul 2011) | 18 lines
+Changed paths:
+   M /trunk/drivers/pdf.c
+
+Use plP_script_scale method to replace idiosyncratic method of
+determining superscript/subscript font sizes and offsets.  The
+test_superscript_subscript.py results are much improved by this
+change.  
+
+However, a vertical offset issue for the pdf device driver that was
+evident before this change still continues; superscript offsets shift
+the entire string downwards (including all text before when "#u" is
+used) while the corresponding result for subscripts (when "#d" is
+used) has no such effect.  I have looked hard for the origin of this
+pdf bug but cannot find the source of it.  The superscript and
+subscript paths are absolutely identical in the code except for the
+numerical calculation of dev->yOffset.  So why does the correct offset
+from the baseline being calculated in both cases affect the overall
+baseline (including the unsuperscripted/unsubscripted _prior_ parts of
+the string) in the superscript case but not in the subscript case?
+
+
+------------------------------------------------------------------------
+r11803 | airwin | 2011-07-10 13:08:00 -0400 (Sun, 10 Jul 2011) | 6 lines
+Changed paths:
+   M /trunk/examples/python/test_superscript_subscript.py
+
+Use more regular vertical alignment of strings.
+
+Add some fiducial marks for the second page that clearly demonstrate vertical
+offset problems with the pdf device driver.
+
+
+------------------------------------------------------------------------
+r11802 | airwin | 2011-07-09 19:11:55 -0400 (Sat, 09 Jul 2011) | 2 lines
+Changed paths:
+   M /trunk/doc/docbook/src/api.xml
+
+State that plspage and plgpage are also used in example 14.
+
+------------------------------------------------------------------------
+r11801 | airwin | 2011-07-08 21:51:34 -0400 (Fri, 08 Jul 2011) | 2 lines
+Changed paths:
+   M /trunk/drivers/svg.c
+
+Oops.  Style previous commit.
+
+------------------------------------------------------------------------
+r11800 | airwin | 2011-07-08 21:49:33 -0400 (Fri, 08 Jul 2011) | 6 lines
+Changed paths:
+   M /trunk/drivers/svg.c
+
+Use plP_script_scale method to replace idiosyncratic method of
+determining superscript/subscript font sizes and offsets.  The
+test_superscript_subscript.py results are much improved by this
+change.
+
+
+------------------------------------------------------------------------
+r11799 | airwin | 2011-07-07 16:56:29 -0400 (Thu, 07 Jul 2011) | 9 lines
+Changed paths:
+   M /trunk/drivers/cairo.c
+
+Use plP_script_scale to calculate superscript/subscript font scale
+factors and vertical offsets.  The results are identical with the
+previous version of this code that made these calculations
+independently of plP_script_scale.  Note plP_script_scale arguments
+must be stored externally, and for the alt_unicode approach this
+storage is necessarily in aStream to preserve the values of these
+arguments between plP_script_scale calls.
+
+
+------------------------------------------------------------------------
+r11798 | hbabcock | 2011-07-06 13:28:53 -0400 (Wed, 06 Jul 2011) | 1 line
+Changed paths:
+   M /trunk/drivers/cairo.c
+
+Fix offset bug in the standard unicode pathway. The problem was that aStream->fontSize is only set in the alt-unicode pathway so the local variable fontSize should be used instead for calculating the text Y offset.
+------------------------------------------------------------------------
+r11797 | airwin | 2011-07-06 01:48:58 -0400 (Wed, 06 Jul 2011) | 6 lines
+Changed paths:
+   M /trunk/drivers/ps.c
+
+Use plP_script_scale method to replace idiosyncratic method of
+determining superscript/subscript font sizes and offsets.  The
+test_superscript_subscript.py results are much improved by this
+change.
+
+
+------------------------------------------------------------------------
+r11796 | airwin | 2011-07-06 00:50:37 -0400 (Wed, 06 Jul 2011) | 5 lines
+Changed paths:
+   M /trunk/bindings/qt_gui/plqt.cpp
+
+Bug fix.  Restore currentFontScale default value that was
+inadvertently removed by the last commit.  This bug affected text for
+a number of qt devices.
+
+
+------------------------------------------------------------------------
+r11795 | airwin | 2011-07-05 22:28:50 -0400 (Tue, 05 Jul 2011) | 8 lines
+Changed paths:
+   M /trunk/drivers/psttf.cc
+   M /trunk/scripts/style_source.sh
+
+Include drivers/psttf.cc in list of C++ source files to style.  (This
+file was inadvertently left out from the original list of source files
+to style.)
+
+Many changes were introduced by the styling of psttfc.cc, but the
+results are the same as for the previous unstyled file.
+
+
+------------------------------------------------------------------------
+r11794 | airwin | 2011-07-05 22:23:16 -0400 (Tue, 05 Jul 2011) | 2 lines
+Changed paths:
+   M /trunk/src/plcore.c
+
+Style previous commit.
+
+------------------------------------------------------------------------
+r11793 | hbabcock | 2011-07-05 22:20:57 -0400 (Tue, 05 Jul 2011) | 1 line
+Changed paths:
+   M /trunk/drivers/cairo.c
+
+Attempt to fix vertical offset problems with strings that contain sub/superscripts. Add string length functionality to the -classic- text rendering path so that it again renders text properly. However the -classic- path is still not completely fixed as there still seems to a offset problem.
+------------------------------------------------------------------------
+r11792 | airwin | 2011-07-05 22:14:40 -0400 (Tue, 05 Jul 2011) | 6 lines
+Changed paths:
+   M /trunk/drivers/psttf.cc
+
+Use plP_script_scale method to replace idiosyncratic method of
+determining superscript/subscript font sizes and offsets.  The
+test_superscript_subscript.py results are much improved by this
+change.
+
+
+------------------------------------------------------------------------
+r11791 | airwin | 2011-07-05 15:06:12 -0400 (Tue, 05 Jul 2011) | 5 lines
+Changed paths:
+   M /trunk/bindings/qt_gui/plqt.cpp
+
+Replace idiosyncratic method of calculating superscript/subscript
+offsets and font sizes with standard (plP_script_scale) method.
+The test_superscript_subscript.py results are much improved by this
+change.
+
+------------------------------------------------------------------------
+r11790 | airwin | 2011-07-05 13:36:24 -0400 (Tue, 05 Jul 2011) | 5 lines
+Changed paths:
+   M /trunk/drivers/cairo.c
+
+Replace generic "smaller" sizes for each superscript/subscript level
+by exact calculated font sizes appropriate for that level.
+
+Some code cleanup and improved commentary.
+
+------------------------------------------------------------------------
+r11789 | airwin | 2011-07-05 13:18:50 -0400 (Tue, 05 Jul 2011) | 8 lines
+Changed paths:
+   M /trunk/examples/python/test_superscript_subscript.py
+
+serif ==> italic to distinguish visual results more clearly when a font
+is changed in the middle of a superscript/subscript level.
+
+Note that -dev qtwidget changes to the italic font at the correct
+point in the string.  -dev xcairo changes to the italic font for the
+_entire_ string (which demonstrates a current cairo bug).
+
+
+------------------------------------------------------------------------
+r11788 | hbabcock | 2011-07-05 13:03:51 -0400 (Tue, 05 Jul 2011) | 1 line
+Changed paths:
+   M /trunk/src/plcore.c
+
+Fix yet another bug in the alternate unicode pathway which was the source of strange font choices in strings that contained more than one font.
+------------------------------------------------------------------------
+r11787 | airwin | 2011-07-04 16:25:58 -0400 (Mon, 04 Jul 2011) | 5 lines
+Changed paths:
+   M /trunk/examples/python/test_superscript_subscript.py
+
+Change to a serif font in the middle of non-zero superstring/substring
+levels to demonstrate more clearly the font-changing issue with the
+cairo devices (the font is changed for the whole string and not
+just the part beyond the FCI command).
+
+------------------------------------------------------------------------
+r11786 | airwin | 2011-07-04 16:22:44 -0400 (Mon, 04 Jul 2011) | 11 lines
+Changed paths:
+   M /trunk/drivers/cairo.c
+
+Fix obvious bug in order in which superscript/subscript level is
+restored for open_span_tag.  This fixes the cairo device driver
+vertical offset issues for test_superscript_subscript.py where FCI
+commands occur at non-zero superscript/subscript levels.
+
+However, this test case also shows an additional issue still remains for
+the cairo devices.  If you change font in the middle of a
+superscript/subscript, it applies to the whole string, not just
+that part of the string beyond where the font change is applied.
+
+
+------------------------------------------------------------------------
+r11785 | airwin | 2011-07-04 15:54:19 -0400 (Mon, 04 Jul 2011) | 2 lines
+Changed paths:
+   M /trunk/include/plplotP.h
+   M /trunk/src/plsym.c
+
+Style previous changes.
+
+------------------------------------------------------------------------
+r11784 | airwin | 2011-07-04 15:37:47 -0400 (Mon, 04 Jul 2011) | 7 lines
+Changed paths:
+   M /trunk/examples/python/test_superscript_subscript.py
+
+Include an FCI command (to change to bold face) in one superscript and
+one subscript test case.  This works correctly for qt devices, but
+fails to give good results for the cairo devices demonstrating a bug
+for those devices for this corner case (FCI and non-zero
+superscript/subscript level).
+
+
+------------------------------------------------------------------------
+r11783 | airwin | 2011-07-03 13:06:18 -0400 (Sun, 03 Jul 2011) | 18 lines
+Changed paths:
+   M /trunk/include/plplotP.h
+   M /trunk/src/plsym.c
+
+Implement a private plP_script_scale function that calculates scale
+values for font size and magnitude of the vertical offset associated
+with superscripts/subscripts.  Currently, the font scale and offset
+calculations associated with superscripts and subscripts are done in
+an idiosyncratic (and often incorrect) way in various locations within
+the core code and drivers.  plP_script_scale is designed to
+consolidate those bits of logic in one place where the calculation is
+done correctly.
+
+As a first use of this new logic, use plP_script_scale for the Hershey
+font superscripts/subscripts. plP_script_scale produces completely
+symmetrical vertical offsets between superscripts and subscripts.
+Therefore, this change corrects a very long-standing error (since
+before 1993) in the vertical offset of the Hershey subscripts which
+previously were not consistent with the magnitude of the vertical
+offsets of the superscripts. 
+
+
+------------------------------------------------------------------------
+r11782 | airwin | 2011-07-01 02:27:21 -0400 (Fri, 01 Jul 2011) | 3 lines
+Changed paths:
+   M /trunk/examples/python/CMakeLists.txt
+   A /trunk/examples/python/test_style.py
+
+Implement test_style.py to show all possible line patterns and area fill
+patterns for display using pllegend.
+
+------------------------------------------------------------------------
+r11781 | airwin | 2011-06-22 16:45:39 -0400 (Wed, 22 Jun 2011) | 20 lines
+Changed paths:
+   M /trunk/drivers/cairo.c
+
+I have discovered experimentally that contrary to the documentation at
+http://developer.gnome.org/pango/unstable/PangoMarkupFormat.html, rise 
+is in units of 1024th of a point (i.e., a real distance unit) rather
+than in units of the (variable) font size.  Implement a correction the 
+accounts properly for the real units of rise. The result is the actual
+superscript/subscript vertical offsets that are rendered by cairo
+devices now is (correctly) proportional to the font size (as opposed 
+to being fixed in size before).  This correction solves the second
+issue noted in the previous commit.
+
+The cairo and PLplot character coordinate system have vertical zero
+points of respectively the baseline of the fonts and the middle of the
+vertical dimension of the fonts. This mismatch in coordinate systems
+means the superscript/subscript offsets appear asymmetrical for cairo
+devices compared to -dev xwin (i.e., issue 1 noted in the previous
+commit message).  An approximation for this coordinate shift has been
+implemented, and the asymmetry appears to be greatly reduced as a
+result.
+
+
+------------------------------------------------------------------------
+r11780 | airwin | 2011-06-22 15:40:38 -0400 (Wed, 22 Jun 2011) | 2 lines
+Changed paths:
+   M /trunk/examples/python/test_superscript_subscript.py
+
+Tweak whitespace.
+
+------------------------------------------------------------------------
+r11779 | airwin | 2011-06-21 23:33:31 -0400 (Tue, 21 Jun 2011) | 10 lines
+Changed paths:
+   M /trunk/examples/python/test_superscript_subscript.py
+
+Implement an additional test which superimposes exponent data you
+obtain with badly scaled axes for plbox with similar (using an
+exponent of -100 rather than -200) exponent data mimicked with direct
+calls to plmtex.  This test shows complete agreement in exponent
+offsets (and good values for those offsets) between the plbox and
+direct calls to plmtex for -dev xwin and -dev xcairo.  Thus, it does
+not verify the bad vertical plbox exponent offsets that you get for
+example 33 using -dev xcairo (but not -dev xwin) 
+
+
+------------------------------------------------------------------------
+r11778 | airwin | 2011-06-21 20:46:58 -0400 (Tue, 21 Jun 2011) | 25 lines
+Changed paths:
+   M /trunk/drivers/cairo.c
+
+Better tuning of superscript/subscript offsets to match algorithm used
+for plstr code in plsym.c that is used for Hershey fonts (e.g., -dev
+xwin).
+
+However, there are still at least two superscript/subscript bugs for
+the cairo devices.
+
+1. test_superscript_subscript.py results show that the vertical
+   offsets for superscripts with -dev xcairo largely match those from
+   -dev xwin.  This is expected because it was those offsets I used to
+   tune the cairo device driver.  However, the subscript vertical
+   offsets are slightly too large in magnitude, i.e., the subscript
+   is placed too low compared to the corresponding superscript.
+   Off by 1 error for subscripts?
+   
+2. The plbox axis exponents that occur for badly scaled axis
+coordinates are written using plmtex.  I mimic those plmtex calls with
+test_superscript_subscript.py.  As expected, the exponent strings from
+plbox and test_superscript_subscript.py with -dev xwin have the same
+vertical offsets. However, the -dev xcairo results for the two cases
+are quite different with the plbox result having much too large a
+vertical offset for the superscript.  Starting superscript offset
+level is off by 1 for plbox?
+
+
+------------------------------------------------------------------------
+r11777 | airwin | 2011-06-21 20:26:12 -0400 (Tue, 21 Jun 2011) | 7 lines
+Changed paths:
+   M /trunk/examples/python/test_superscript_subscript.py
+
+Add the equivalent plmtex superscript and subscript tests to the
+previous plptex tests.
+
+Add four specific plmtex tests that mimic the exponents that occur for
+badly scaled plbox axes.
+
+
+------------------------------------------------------------------------
+r11776 | airwin | 2011-06-21 14:08:09 -0400 (Tue, 21 Jun 2011) | 10 lines
+Changed paths:
+   M /trunk/examples/c/x33c.c
+   M /trunk/src/plbox.c
+
+Update bounding box to include both vertical and horizontal components
+of numerical labelling exponents for horizontal axes.  Example 33
+confirms this improvement is working properly.  This completes the
+bounding box calculation for horizontal axes of plbox.
+
+ToDo: Need to implement similar functionality for vertical axes of
+plbox, horizontal and vertical axes of custom labels, and overall
+colorbar label. 
+
+
+------------------------------------------------------------------------
+r11775 | airwin | 2011-06-21 12:43:35 -0400 (Tue, 21 Jun 2011) | 9 lines
+Changed paths:
+   M /trunk/src/plbox.c
+   M /trunk/src/plleg...
 
[truncated message content]

Thread: [Plplot-cvs] SF.net SVN: plplot:[11831] trunk/ChangeLog.release

Cross-platform, scientific graphics plotting library

plplot-cvs