This file is designed to keep track of various problem fixes and new features being worked on for the current PLplot CVS MAIN HEAD version and who is taking responsibility for following up on each item. The items are divided into Essential and "Would be nice" (i.e., release critical versus not) with each of those categories further classified into minor or major effort required. N.B. at the end we have an additional section on problems/issues associated with the new unicode-aware fonts. Essential, major effort required. ******************************** Essential, minor effort required. ******************************** * PLplot documentation of python, java, etc., interfaces. Alan "Would be nice", major effort required. *************************************** * Fix bad-looking surface contours for plsurf. The problem goes away if the plP_fill call is commented out in the normal series of alternating calls to plP_fill and plline3 that are used to generate the surface and the surface contours. So I conclude there is some inconsistency between the edge of the triangles as defined for the surface contours plotted by plline3 and the edge of the triangles being filled. Low resolution devices have worse problems than high resolution devices. Surface contours look bad for png, better (but still not acceptable for publication) for psc, better still for xwin, and I could not find any problems at all for tk. * Move from DSSL based docbook stylesheets to XSL. Rafael, as soon as the XSL results look as good (pdf currently a problem) as the DSSL results. * Maurice's idea of plsopt and plsdopt so that our API doesn't keep proliferating. Example, instead of plenv0 do: plsopt("plenv_advance", "0"); plenv(...) * -the tk drivers ignore plconfig.tcl if it is in the user directory ~/tcl and a tclIndex file is generated. "auto_path" is correctly initialized to ~/tcl, pwd/tcl, etc, but plconfig is not executed. Reported by Joao. Status? * Maurice plans to finish his changes to implement handling strings as strings in the metafile format. * Geoffrey plans to at least evaluate remaining fidelity problems caused by our 16-bit integer approach. * Put in index limits for the plot3d API's so they can handle non-rectangular x,y regions. This has already been done for plsurf and needs to be extended to plot3d. * Put in variation of "defined" callback function with user-defined data for plshade and plshades. Rafael? * Put in "defined" callback function with user-defined data for plcont, plot3d, and plsurf. This will supersede the non-rectangular x,y regions change already done for plsurf and planned for plot3d, but this programming is harder than those previous or planned changes so it will probably take longer. Rafael? "Would be nice", minor effort required. *************************************** * 2003-09-20: The generated info files should have their root name changed from plplotdoc (plplotdoc.info, etc.) to plplot$version like the rest of the generated files (plplot-5.2.1.cvs.20030915.dvi, etc.) Note this change would require that packagers' scripts that execute install-info would need a change to use the versioned name, but this should be straightforward to do both for RedHat and Debian packages. * 2003-09-20: References are not there in info bibliography section (chapter 14). According to Rafael this problem is caused by docbook2x not currently supporting the and tags so we will have to wait for that to get sorted out. * plplot_general list 2003-08-26: Arjen Markus "I have found that with the X Window driver the values returned by plgspa() are/can be too small by a factor of approximately 1.4." This should be followed up by an attempt to confirm the bug, and if it is there, fix it. * Error messages for when the extended search fails in tclAPI.c. This is required when running in tcl-only mode with no plframe or stack trace. * As part of extended search try the install location. We know exactly where that is relative to our font data, and we know exactly where the font data are installed so we know the absolute path to our tcl install location. Then, the only way such an extended search would fail is if the packagers messed with our tcl install location. * Use some well-recognized pdf verification tool (similar in spirit to the w3c verification tests for web sites) to be sure our results conform to the published pdf standards. Currently there is no idea whether such a pdf verification tool exists, but acroread is buggy on Linux so that is not good enough as a verification tool. Collection of problems/issues with new unicode-aware fonts as of 2005-03-21. All these are in the would-be-nice category with some/all taking major effort to resolve. **************************************************************************** I. Font problems associated with plfreetype.c which is called by gd.c which implements the png, gif, or jpeg devices. (These issues were discovered using -dev png). * The default -drvopt smooth=1 option does not work (lots of cmap0 index errors) if there is any user change to cmap0. This problem is triggered if you attempt to run example 2 with -dev png. Andrew Roach thinks he knows a small redesign that will fix it. * There are still problems with vertical alignment of symbols for plfreetype.c. See results from examples/c/x06c -dev png -o test.png where on line 100 the characters move up or down depending on whether they have an ascender, descender or neither. There are similar problems for example 7. N.B. This is only a problem for plpoin (example 6) and plsym (example 7), but not a problem for plmtex. So there is probably some issue with the way plhrsh (called by plpoin and plsym) sets up the call to the code in plfreetype.c that is not present for plmtex. * The above problem may be related to the xform setting in plhrsh (plsym.c) which is a kludge. if you define TEST_FOR_MISSING_GLYPHS, it quits working (all the symbols are twisted by 90 deg). * -drvopt 24bit problems. If this option is turned on (and only in this case), Example 9 and 22 second and subsequent pages have rectangular viewports with no box or tick marks. Furthermore, with Example 19 the second and 3rd pages turn black. If you use -bg FFFFFF, you can see the problem is a cmap0 one, somehow 24bit turns the default red colour into black which doesn't show against the normal black background, but which does show against a white background as with -bg FFFFFF. * text clipping does not work (see example 9). There is a clipping method in libgd, but the preferred solution (since it will apply to every device driver that uses plfreetype.c) is to use the libfreetype clipping method which Andrew Ross found in the freetype documentation. * We need to build in fontconfig support into plfreetype.c for truetype fonts. This well-known library by Keith Packard is available for both Linux and windows and should allow truetype font selection by name rather than by filename, and font substitution on both Linux and windows if the requested named font is not available. II. Font problems associated with ps.c and either -dev psc or -dev ps. * The overall size of the text is systematically smaller than for Hershey or TrueType fonts. III. Other font problems/issues. * We should enable additional device drivers for unicode. In particular, our two most heavily used interactive device drivers, xwin and tk, should give some outstanding looking results once they are unicode-enabled. I nominate Maurice to make this happen.... :-)