Gnumeric 1.3.1 aka Polished Weasel

[mortenw gnome org (Morten Welinder)]

Gnumeric 1.3.1 aka Polished Weasel is out.  This release is a development
release and the usual warnings about potential crashes and data-mangling
apply.

Fixes and improvements are everywhere.  Some of the most noteworthy
are: improved xls graph import, initial support for rich text, 
better xls evaluation compatibility for operators, and improved
accuracy for lots of statistical and financial functions.

A Win32 version is being worked on.  More patches are needed, but
volunteers have gotten it to actually start.

Thanks to all our contributors!

Morten
on behalf of the team


Andreas:
        http://bugzilla.gnome.org/show_bug.cgi?id=142219
        * Handle borders in latex export correctly
        http://bugzilla.gnome.org/show_bug.cgi?id=142836
        * Save current sheet as csv (rather than first sheet)
        http://bugzilla.gnome.org/show_bug.cgi?id=140820
        * Fix default colour handling in manage sheet dialog
        * Add interface to allow the visibility of sheets to be toggled
        http://bugzilla.gnome.org/show_bug.cgi?id=142700
        * Make sure zoom-box shows current zoom factor
        * Improve `Descriptive Statistics' tool
        * Improve `One-Factor ANOVA' tool
        * Improve `Correlations' tool
        * Improve `Covariances' tool
        * Add CLIPBOARD over PRIMARY choice to preference dialog

Christian Neumair:
        * Put delete actions from edit main menu into their own submenu.
        * HIGify some dialogs when it comes to spacing.

Emmanuel Pacaud:
        http://bugzilla.gnome.org/show_bug.cgi?id=142212
        * In a line graph, don't use '0' if Y value is missing.
        * fix a double draw of chart childs.

Jean Brefort:
        * Add error bars in bars, columns, lines and areas plots.

Jody:
        * Edit and display rich text in cells (no persistence yet)
        * Make the last dependency above gtk optional to enable a win32 build
        * Major re-org of xls string import to handle rich text
        * Improve xls import of old < xl95 variants
        * Fix xls chart import for value dimensions
        * Fix autoformat dialog
        * Fix db functions
        * Pango uses _byte_ offsets for attributes.  Fix everything
        * Fix xls import of chart background and outline
        * Support richtext during autofill
        * A huge performance improvement when scrolling images
        http://bugzilla.gnome.org/show_bug.cgi?id=136363
        * Fix font handling in chart labels and titles
        http://bugzilla.gnome.org/show_bug.cgi?id=140671
        * Fix crash when undoing a cut from a different workbook
        * Fix MS XL crash when there are too many strings
        * MS XL import for chart gradient backgrounds
        * Support series without legend entries, including xls import
        * Started on an OOo exporter
        http://bugzilla.gnome.org/show_bug.cgi?id=103054
        * Tweak XL import and export for named expressions / addins
        http://bugzilla.gnome.org/show_bug.cgi?id=142474
        * Use first rather than last conditional value fmt as catch-all
        * Implicit iteration for operators
        * Make the unary operators XL compatible
        * basic xls export for charts
        * start work on OOo export
        * Skip ticks/labels for discrete axes
        * Drag files from nautilus -> gnumeric
        http://bugzilla.gnome.org/show_bug.cgi?id=143577
        * fix grid printing
        * fix merged cell printing and diagonal borders
        * Allow themed series with gradients
        * Fix application prop dialog
        * Support repeat for several commands
        http://bugzilla.gnome.org/show_bug.cgi?id=147756
        * Ancient autofill leak on constant string fills

Morten:
        * Fix crash related to database functions (#138600).  [backported]
        * Fix [h]:mm precision (#138182).  [backported]
        * Make it possible to hide toolbars.
        * Make it possible to hide the statusbar.
        * A few multihead fixes.
        * Upgrade to R's development version.
        * Add cumulative distribution to HYPGEOMDIST.
        * Fix pgeom for small x.  (Fixes accuracy of GEOMDIST.)
        * Fix pbinom for very large n.  (Fixes accuracy of BINOMDIST.)
        * Replace R's pbeta function.  (Fixes accuracy of BETADIST,
          BETAINV, FINV.)
        * Replace R's pcauchy function .  (Fixes accuracy of CAUCHY.)
        * Replace R's pf function.  (Fixes accuracy of FDIST, FTEST, F-test
          tool, regression tool, ANOVA tools..)
        * Replace R's phyper function.  (Fixes accuracy and evaluation speed
          of HYPGEOMDIST.)
        * Fix accuracy of our versions of various long-double functions
          that Solaris lacks.  (Fixes, for example, ERF.)
        * Improve accuracy of FISHER.
        * Improve accuracy of EFFECT, NOMINAL, MIRR, IRR, XNPV, NPV, PMT,
          RATE, PV, FV, IPMT, PPMT for small rates.
        * Fix accuracy of ZTEST.
        * Improve prevision of the bessel functions.
        * Fix validation crash due to excel import.  [backported]
        * Fix custom validation export.  [backported]
        * New functions EXPM1 and LN1P.
        * Fix image-inside-xls problem.  [backported]
        * Ignore more spaces in expression entry.
        * Improve error handling in xbase plugin.  [backported]
        * Fix a weird IF semantics where the argument count including
          empties matters.
        * Fix parser problem with single empty arg after expression.  [backported]
        * Avoid keeping passwords in memory.  (Will need gtk+ fix to be really
          effective.)
        * Switch to keeping URIs, not filenames, inside and in history.
        * Implement copy-right (Ctrl-R) and copy-down (Ctrl-D).

        (For those worrying over the pile of accuracy fixes above it should
        be noted that prior versions of Gnumeric were already pretty good,
        certainly far better than Excel.  Most of the fixes above are for
        extreme-case arguments where most other software packages will give
        totally incorrect answers and/or take extremely long time to
        produce any answer at all.  We are slowly feeding some of these
        fixes back to, say, the R Project.)