Bugzilla Bug 238818

Here's a new minor bugfix version, which also produces standard compliant SVG
1.1.

NEWS
Version 2.5.1 released, 7/2008.  [Includes libplot 4.3, interface age=2.]

package: license for the package (and for libplot) upgraded to GNU GPL v3.
libplot: SVG output format slightly tweaked, to ensure compatability with
         SVG 1.1 syntax, as tested for by validator.w3.org.
         Note: a few SVG editing tools, such as inkscape, currently fail
         to agree with libplot's SVG semantics, in positioning text.
         That appears to be their fault (they do not comply with the
         handling of reflected text, as specified in the SVG 1.1 standard).
libplot: Fixed a bug in bounding box computatations when triangular 
         line-joins are used.
executables: Changed nearly all warning and error messages to complete
         sentences, to smooth the upcoming transition to gettext.

COMPAT
Release 2.5.1, dated 7/2008
---------------------------

This release fixes a few bugs.  However, there are persistent problems with
text-strings in the SVG output format.  Many applications that can view or
edit SVG files (in particular, inkscape) do not correctly implement the SVG
1.1 specification.  So, they may position text strings incorrectly, and
even position incorrectly individual characters within a text string.

If you encounter this problem, the fix is to specify a stroked ("Hershey")
font, such as HersheySerif or HersheySerif-Bold.  A list of available
fonts, in SVG output, is obtained by doing `graph -Tsvg --help-fonts'.
Each character in any Hershey font is made up of stroked line segments, and
most SVG viewers can display line segments correctly.

This problem will be worked around in future releases of the SVG output
driver.

Created an attachment (id=166553) [edit]
plotutils-2.5.1-fix-tests.patch

There are two failing tests.

One of the failured is simply because the version string wasn't updated on the
tests.

The other is kind of outdatedness too... I assumed that the output produced by
plotutils is correct (it passes SVG validator) instead of the hardcoded one
(doesn't validate).

I'm not sure the second part of the patch is correct. I've sent it upstream and
I'm waiting for their answer.

Santiago have you received any answer from upstream? Also since some open bugs
are already fixed but nobody cared to close them I suppose nobody looks at this
package at the moment (printing, please correct me if I'm wrong), so may be
you'll take care of this package?

Created an attachment (id=171104) [edit]
plotutils-2.5.1-rangecheck.patch

Updated patch (in unified format) and added comments about reason and source.

Created an attachment (id=171106) [edit]
plotutils-2.5.1.ebuild

Updated ebuild.

Peter, I haven't received any answer from upstream.

I'm not interested in taking over maintainership of this since I don't really
use this package. However, I can take care of the current bugs.

If printing team has no objections, I'll commit this ebuild and ask maintainers
of reverse dependencies to test it (as well as review the use of 'plotutils'
USE flag), and then ask for stabilization since >=2.5 is needed for gcc-4.3.1.

Santiago, silence gives consent. ;)

Done ;-)