dia_improved

Fork

path: fix Difference sometimes not closing the path

distance_bez_shape_point() was miscalculating the distance of
non-closed path making selection harder. Now handled because
such path can be created by other means, too.

_curve_from_segment() did not correctly flip a line-to, the
remaining changes should simply increase readability (typos,
documentation updates)

unrelated: stdpath_get_object_menu() make "Show Control Lines"
a toggle menu.

Bug 729199 - Set xml:space="preserve" to align text starting with spaces

To minimize impact of this deprecated attribute we set it as locally as
possible and only if it is needed at all. Works for Inkscape, Firefox and
Chrome but not for IE11.

Bug 735430 - vdx: Box rounding support for export and import

VDX 'Rounding' is basically the same as Dia corner_radius, so
draw_rounded_rect() was really easy to implement. While at it
remove all superfluous prototypes instead of adding a new one.

The import has to be a little more elaborated, because corner_radius
can only be handled with "Standard - Box", but VDX has no rectangle.
So we check the polygon for being rectangular and if so apply the
rounding.

Checked with ellipse.dia round-trip as well as with Visio Viewer 2010.

Also included some minor line style/caps tweaking.

layer-widget: Add edit name button to button row

This makes the layer list slightly wider, so it is loosing it's
scroll-bar for most diagrams I looked at.

dox: improved documentation for renderers and importers

The DiaRenderer interface is now completely documented with renderer
implementors in mind. To support sub-grouping in the documentation some
code needed to be moved around, but there is no functional change.

Also more importers are included with the Doxygen generated documentation.

vdx: fix position of bottom aligned text

Not shifting by height makes text position right for text_tests.vdx
Doing no shift for the other VerticalAlign screws them ...

Bug 735430 - Implement VDXRenderer::fill_arc()

by copying draw_arc() implementation, changing to fill and adding two line
segments. To get even nicer appearance of boxes with rounded corners an
implementation of VDXRenderer::draw_rounded_rect() would be needed. But
that's an enhancement left for another time.

Bug 606377 follow-up: beautify filter names and fix help for --list-filters

Now that dia --list-filters shows all the available export filter strings
replace some abbreviations with the full name and shorten the longest
filter name to fit typical terminal width.

dox: [warningectomy] missing parameter documentation ...

- add plug-ins/pdf for documentation extraction
- remove second occurrence of plug-ins/svg
- fix doxygen complaints about missing parameter documentation
- fix a bug in export-render.py where the parameter was documented,
used in the implementation but not declared
- fix title mismatch for group ObjectConnects

vdx: optimize VDXRenderer::draw_ellipse() to combine fill and stroke

VDX can have fill and/or stroke with a single shape by just setting the
right properties.

vdx: optimize VDXRenderer::draw_polygon() to combine fill and stroke

VDX can have fill and/or stroke with a single shape by just setting the
right properties. Also now stroked polygons are staying polygon rather
than becoming polyline.

Update NEWS from dia-0-97 branch

dia-0.97.3: 2014-09-05

* Fix double free with some SVG rendering (regression from Dia 0.97.2)
* Fixes to cope better with updated versions of Dia's dependencies:
- don't crash at start-up with ABI breaking GLib 2-36
- don't assert in cairo 1.12 with invalid arc parameters
- avoid kerning problems (character overlap) for all Pango versions
- fix image files to be loadable by libpng16
* Backport fixes for some seldom crashes and other annoyances, see:
http://bugzilla.gnome.org/buglist.cgi?product=dia&target_milestone=0.97.3
* Update translations for Brazilian Portuguese, German, Hungarian,
Polish, Romanian, Serbian and Slovenian

Dia 0.97.3 - bug-fix release (NEWS)

dia-0.97.3: 2014-09-05

* Fix double free with some SVG rendering (regression from Dia 0.97.2)
* Fixes to cope better with updated versions of Dia's dependencies:
- don't crash at start-up with ABI breaking GLib 2-36
- don't assert in cairo 1.12 with invalid arc parameters
- avoid kerning problems (character overlap) for all Pango versions
- fix image files to be loadable by libpng16
* Backport fixes for some seldom crashes and other annoyances, see:
http://bugzilla.gnome.org/buglist.cgi?product=dia&target_milestone=0.97.3
* Update translations for Brazilian Portuguese, German, Hungarian,
Polish, Romanian, Serbian and Slovenian

xfig-import: don't create a group for a single object

it's just superfluous and annoying during round-trip.

[scan-build] Result of operation is garbage or undefined (invalid)

Help scan-build to understand constraints on handles, too.

The left operand of '-' is a garbage value
File: /home/hb/from-git/dia/tests/test-objects.c
Line: 244

[warningectomy] no return statement in function returning non-void

export_png.c: In function 'export_png':
export_png.c:379:1: warning: no return statement in function returning non-void

[warningectomy] format '%d' expects argument of type 'int'

message.c: In function 'dia_log_message':
message.c:334:3: warning: format '%d' expects argument of type 'int', but argument 4 has type 'gulong' [-Wformat=]
g_message ("%02d:%02d:%02d.%03d - %s", h, m, s, ms/1000, log);
^
message.c:334:3: warning: format '%d' expects argument of type 'int', but argument 5 has type 'gulong' [-Wformat=]
message.c:334:3: warning: format '%d' expects argument of type 'int', but argument 6 has type 'gulong' [-Wformat=]
message.c:334:3: warning: format '%d' expects argument of type 'int', but argument 7 has type 'long unsigned int' [-Wformat=]

[scan-build] Dead increment (harmless)

Value stored to 'h' is never read
File: /home/hb/from-git/dia/objects/UML/node.c
Line: 304
-> removed

Value stored to 'y' is never read
File: /home/hb/from-git/dia/objects/UML/object.c
Line: 314
-> removed

[scan-build] Dead initialization (harmless)

Value stored to 'extra' during its initialization is never read
File: /home/hb/from-git/dia/lib/prop_sdarray_widget.c
Line: 524
-> removed variable and it's initialization

Value stored to 'elem' during its initialization is never read
File: /home/hb/from-git/dia/objects/Misc/n_gon.c
Line: 191
-> use it a little below, rather than calculating the pointer once more

Value stored to 'otype' during its initialization is never read
File: /home/hb/from-git/dia/plug-ins/svg/svg-import.c
Line: 893
-> remove intialization

Value stored to 'renderer' during its initialization is never read
File: /home/hb/from-git/dia/plug-ins/wpg/wpg.c
Line: 772
-> move intialization into the code branch using the variable

Value stored to 'hBrOld' during its initialization is never read
File: /home/hb/from-git/dia/plug-ins/wmf/wmf.cpp
Line: 1177
-> remove all the assignments to hBrOld and it's declaration

Value stored to 'renderer' during its initialization is never read
File: /home/hb/from-git/dia/plug-ins/wmf/wmf.cpp
Line: 1064
-> move intialization into the code branch using the variable

Value stored to 'wmf_renderer' during its initialization is never read
File: /home/hb/from-git/dia/plug-ins/wmf/wmf.cpp
Line: 1232
-> move deinitialization into *_finalize from end_render()

[scan-build] Fix potential memory leak

Value stored to 'renderer' during its initialization is never read
File: /home/hb/from-git/dia/lib/diaimportrenderer.c
Line: 142

A user of DiaImportRenderer which would not call
dia_import_renderer_get_objects() after importing the data would
have leaked all the objects.

[scan-build] Dereference of null pointer (probably not)

Make a superfluous check for Any!=NULL a g_return_val_if_fail()
otherwise the static analysis assumes Any could be NULL in this
function an will later complain about the missing check. But
we are never putting NULL into the list of items.

Access to field 'type' results in a dereference of a null pointer
(loaded from variable 'Any')
File: /home/hb/from-git/dia/plug-ins/vdx/vdx-import.c
Line: 1822

[scan-build] Dereference of null pointer (maybe)

Access to field 'len' results in a dereference of a null pointer
(loaded from field 'pathes')
File: /home/hb/from-git/dia/lib/diapathrenderer.c
Line: 870

[scan-build] Dereference of null pointer (maybe)

Access to field 'len' results in a dereference of a null pointer
(loaded from variable 'p1')
File: /home/hb/from-git/dia/lib/diapathrenderer.c
Line: 887

[scan-build] Dereference of null pointer (maybe)

Access to field 'object' results in a dereference of a null pointer
(loaded from variable 'cp')
File: /home/hb/from-git/dia/lib/autoroute.c
Line: 276

xfig: optimize draw_(ellipse|polygon|rect) to combine stroke and fill

Tested with a modified render-test.dia because the original converted to
.fig make xfig 3.2.5c crash on every linestyle other than solid or dotted.

Also tested via xfig plug-in round-trip. For some reason the new single
objects for ellipse, rect and polygon are still put into a group. But
that's an issue to be tackled in xfig-import.c if at all.

Fix recently observed crash with "UML - Component Feature"

found by make distcheck running test-objects and toggling the
auto-routing via object menu function plus reverting it.

gdb --args .libs/objects ../objects//
[...]
Starting program: [...]/dia/tests/.libs/objects ../objects//
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/i386-linux-gnu/libthread_db.so.1".
/Dia/Objects/UML - Component Feature/Creation: OK
/Dia/Objects/UML - Component Feature/Copy: OK
/Dia/Objects/UML - Component Feature/Movement: OK
/Dia/Objects/UML - Component Feature/Change: OK
/Dia/Objects/UML - Component Feature/MoveHandle: OK
/Dia/Objects/UML - Component Feature/ConnectionPoints: OK
/Dia/Objects/UML - Component Feature/ObjectMenu: ***
Error in `[...]/dia/tests/.libs/objects': double free or corruption (fasttop): 0x0817b1d8 ***

Program received signal SIGABRT, Aborted.
0xb7fdd424 in __kernel_vsyscall ()
(gdb) bt
0 0xb7fdd424 in __kernel_vsyscall ()
1 0xb70dd577 in __GI_raise (sig=sig@entry=6)
at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
2 0xb70e09a3 in __GI_abort () at abort.c:89
3 0xb71186e3 in __libc_message (do_abort=do_abort@entry=1,
fmt=fmt@entry=0xb72157bc "*** Error in `%s': %s: 0x%s ***\n")
at ../sysdeps/posix/libc_fatal.c:175
4 0xb7122bca in malloc_printerr (action=<optimized out>,
str=0xb7215980 "double free or corruption (fasttop)", ptr=0x817b1d8)
at malloc.c:4996
5 0xb712383d in _int_free (av=0xb725b420 <main_arena>, p=<optimized out>,
have_lock=0) at malloc.c:3840
6 0xb740bd30 in g_free (mem=0x817b1d8)
at /build/buildd/glib2.0-2.40.0/./glib/gmem.c:190
7 0xb7f8613b in orthconn_destroy (orth=0x8189f98) at orth_conn.c:582
8 0x0804aef8 in _test_object_menu (user_data=0xb6c99284 <compfeat_type>)
at test-objects.c:571

vdx: adapt <ForeignData/> attributes and content

At least LibreOffice 4.2.4.2 is not happy with the padding
after base64 data, not even with the linebreak following. See:
https://bugs.freedesktop.org/show_bug.cgi?id=83275

The change in lib/prop_pixbuf.c to not put a linebreak into
the base64 data will have effect on other output as well,
but the tag directly following should not be a problem
anywhere.

vdx: write a single page size defintion ...

... Visio Viewer does not care, but LibreOffice does.
Also remove bogus optional attributes from Page tag.

Fix shape.dtd to include rx, ry attributes for rect

found by make distcheck

cgm: add header to Makefile.am (found by make distcheck)