- Inkscape Devel - Inkscape Mailman

NEW: speedups, Outline mode
by bulia byak 13 Nov '05

13 Nov '05

(This and the following messages describe new stuff in CVS and thus in 0.44, not 0.43) Recently I committed a few rendering speedup patches that I've been testing for a few weeks. It's still far from Xara speed, but the difference is noticeable. 1. As you probably noticed, screen redraw is done in chunks (typically horizontal strips). The maximum size of the chunk was 64K, which is pretty small. (Lauris had a strange idea that Sodipodi must run fine on a system with 32Mb RAM.) On the other hand, rendering each chunk causes a complete stroking and uncrossing of the paths touched by the chunk, which is especially expensive for paths with strokes, and even more so for dash patterns. This means the monstrous livarot functions were called a lot more often than necessary. I increased the chunk size to 256K, and that helped a lot. In some critical situations (complex stroked/dashed paths at high zooms) it's now up to 3 times faster; typically at least 10% faster. Node editing is also noticeably snappier because it involves a lot of path rendering. I then tried 1M chunks, but even though this made stroke/path rendering still faster, for some reason it slowed down rendering of files with lots of semitransparent gradients. So in the end, I settled on 256K. The gradient slowness with large chunks needs further investigation. 2. When rendering transparent overlays, Inkscape spends a lot of time calculating NR_PREMUL(c,a) defined as (((c) * (a) + 127)/255). I changed the division to FAST_DIVIDE_BY_255(v) ((((v)<< 8) + (v) + 257) >> 16) borrowed from Mozilla and got about 10% speedup on complex documents with transparency. 3. An error in sp-canvas.cpp caused a complete extra screen redraw after each zoom operation. That is, after you press "+" in a complex drawing, Inkscape redraws, but for some time after that it remains still unresponsive because it does that second redraw (invisibly for you, i.e. nothing changes on the screen). I fixed that, and now zooming is much more pleasant in complex docs. 4. It turned out pretty easy to code an Outline ("wireframe") display mode. In it: - all paths and shapes are shown as inverse (black on light background and vice versa) outlines of constant width (i.e. 1 screen pixel regardless of zoom), without fill; - text is painted by inverse fill color without stroke; - bitmaps are shown as is; - any opacity and gradients are ignored. Since Inkscape's display of path strokes is the slowest part of its renderer, the outline mode is usually not drastically faster than regular mode (usually 10% to 50% faster), and in some cases it may even be slower. However, the value of the outline mode is not only in its speed; it is a good way to get an idea of the structure and objects of your document, and it is convenient for precision node editing and for finding "stray objects". The switch from normal to outline view and back (View > Display mode) works on the fly. (As for a non-AA mode, I have no idea how to do it, and anyway I doubt that it will give a noticeable speedup at all.) I'm also really looking forward to Xara opening its code. I do plan to attempt to borrow their renderer, or parts thereof, into Inkscape, now that I know that ins and outs (but not the guts) of our own renderer fairly well, but I cannot promise anything - this may turn out prohibitively difficult. We'll see. -- bulia byak Inkscape. Draw Freely. http://www.inkscape.org

7 11

New warnings (was Re: inkscape build report)
by Bryce Harrington 13 Nov '05

13 Nov '05

Three new warnings popped up in the build today... Bryce ----- Forwarded message from root <root@...1070...> ----- Date: Sat, 12 Nov 2005 17:21:02 -0800 From: root <root@...1070...> To: inkscape-tester(a)lists.sourceforge.net Subject: [Inkscape-tester] inkscape build report = Analysis Report for inkscape = Report Time: Sat Nov 12 15:54:54 2005 (Run 747) Snapshot: inkscape-20051112-1440.tar.bz2 Status: finished Platform Unpack Config Build Install WRNS ERRS ------------------------------------------------------------------------ debian x86 0 0 0 0 7 0 Test Status Score Info ------------------------------------------------------------------------ sut-1 ?? == Errors == none == Warnings == seltrans.cpp:651: warning: `void sp_sel_trans_handle_click(SPKnot*, unsigned int, void*)' defined but not used sp-item.cpp:151: warning: converting to non-pointer type `double' from NULL sp-item.cpp:152: warning: converting to non-pointer type `double' from NULL display/canvas-arena.cpp:87: warning: invalid access to non-static data member `SPCanvasArenaClass::arena_event' of NULL object display/canvas-arena.cpp:87: warning: (perhaps the `offsetof' macro was used incorrectly) display/sp-canvas.cpp:133: warning: invalid access to non-static data member `SPCanvasItemClass::event' of NULL object display/sp-canvas.cpp:133: warning: (perhaps the `offsetof' macro was used incorrectly) == Environment Summary == ------------------------------------------------------- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php _______________________________________________ Inkscape-tester mailing list Inkscape-tester(a)lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/inkscape-tester ----- End forwarded message -----

1 0

RE: 0.43pre3: Everything is perfect
by James Kilfiger 13 Nov '05

13 Nov '05

> And, if it's not, now is a good time to mention it. Well there're bugs https://sourceforge.net/tracker/?func=detail&atid=604306&aid=1281512&group_… https://sourceforge.net/tracker/?func=detail&atid=604306&aid=1335301&group_… which are gdk bugs that make using inkscape with the mouse of a tablet on win32 pretty difficult. James Kilfiger

1 0

No AR_FLAGS in src/Makefile.in
by Jeremy Y. Meng 12 Nov '05

12 Nov '05

I noticed that there is this line AR = @AR@ in src/Makefile.in, but no AR_FLAGS line in the file. The default value of 'cru' is hardcoded everywhere. Is this a bug or else? -- Jeremy ---------------------------------------------------------------------------

1 0

boolean ops bugs
by unknown＠example.com 12 Nov '05

12 Nov '05

The itch I would most like to scratch is fixing the bugs that cause objects to disappear with boolean ops. I can't believe how often I run into these bugs when I try to do real work in inkscape. (I almost never try to do real work.) I've also noticed that newbies often try boolean ops with simple paths to see the results, in many cases simple paths are the ones that cause boolean ops the most problems. The bugs I witness are similar to the following: http://sourceforge.net/tracker/index.php?func=detail&aid=1056182&group_id=9… http://sourceforge.net/tracker/index.php?func=detail&aid=1240704&group_id=9… http://sourceforge.net/tracker/index.php?func=detail&aid=1121214&group_id=9… Tracking boolean ops down in doxygen I get about this far before I realize that I don't understand anything: http://www.inkscape.org/doc/doxygen/html/splivarot_8cpp.php#689a3eb52238706… http://www.inkscape.org/doc/doxygen/html/classShape.php#280456eade08d59a110… I'm not going to be able to track and squash these bugs on my own. I have no idea what process to follow or what to look for. But I do pledge to assist in any way the person or persons who will offer to help me. Aaron Spike

2 1

Port to SGI IRIX 6.5
by Jeremy Y. Meng 12 Nov '05

12 Nov '05

Dear Inkscape developers, Thank you all for this wonderful program. I love it and have been trying to port it to SGI IRIX 6.5 platform. My workstation is an Octane2 V6 running 6.5.28. My goal is to build Inkscape-20051109-0840 using SGI MIPSpro compiler. I managed to get an executable by patching the source (many many "#ifdef (__sgi) ..."). However it crashed every time shortly after startup, with the following error: > inkscape event_scan: got unknown client message ALERT: Process [inkscape] 775729 generated trap, but has signal 10 held or ignored epc 0xfa46a14 ra 0xfa46244 badvaddr 0xffffffffff396d98 process has been terminated to prevent infinite loop Bus error The interesting part is that I could run inkscape remotely from a windows machine via Cygwin/X server and play with it for a while (until it randomly crashed). Does this have anything to do with the X visuals? The Octane V6 has a highest 30-bit visual. Then I tried again using gcc 3.4. This time I only need to modify a few #define's regarding isFinite() and round(). But the disaster was when making the inkscape binary. There are tons of errors like a few listed below. I don't know too much C++. I am hoping my patch didn't introduce any bugs into Inkscape. Anyone interested please take a look at my patch and maybe give some insights? --- begin pasting error msg --- gmake all-recursive gmake[1]: Entering directory `/usr/people/voidfoo/devel/inkscape/inkscape_gcc/build_gcc' Making all in src gmake[2]: Entering directory `/usr/people/voidfoo/devel/inkscape/inkscape_gcc/build_gcc/src' /bin/sh ../libtool --mode=link g++ -Wall -W -Wpointer-arith -Wcast-align -Wsign-compare -Woverloaded-virtual -Wswitch -Wno-unused-parameter -g -O2 -Wa,-xgot -L/usr/nekoware/lib -o inkscape --export-dynamic main.o libinkpre.a application/libinkapp.a ui/dialog/libuidialog.a dialogs/libspdialogs.a dom/libdom.a jabber_whiteboard/libjabber_whiteboard.a trace/libtrace.a svg/libspsvg.a widgets/libspwidgets.a display/libspdisplay.a helper/libspchelp.a libcroco/libcroco.a libnrtype/libnrtype.a libnr/libnr.a libavoid/libavoid.a livarot/libvarot.a ui/view/libuiview.a ui/libui.a ui/widget/libuiwidget.a extension/libextension.a extension/implementation/libimplementation.a extension/internal/libinternal.a extension/script/libscript.a xml/libspxml.a util/libinkutil.a io/libio.a inkjar/libinkjar.a libinkpost.a debug/libinkdebug.a -L/usr/nekoware/lib -L/usr/nekoware//lib -lgtkmm-2.4 -lgdkmm-2.4 -latkmm-1.6 -lpangomm-1.4 -lglibmm-2.4 -lgtk-x11-2.0 -lgdk-x11-2.0 -latk-1.0 -lgdk_pixbuf-2.0 -lpangocairo-1.0 -lpango-1.0 -lcairo -lgobject-2.0 -lgmodule-2.0 -lxslt -lxml2 -lz -liconv -lm -lsigc-2.0 -lgthread-2.0 -lpthread -lglib-2.0 -lintl -lpng -lz -lpopt -L/usr/nekoware/lib -lgnomevfs-2 -lbonobo-2 -lgconf-2 -lbonobo-activation -lORBit-2 -lm -lgmodule-2.0 -lgthread-2.0 -lpthread -lglib-2.0 -lintl -rpath /usr/nekoware/lib -L/usr/nekoware/lib -lXft -lfreetype -lz -lXrender -lX11 -lfontconfig -L/usr/nekoware/lib -lpangoft2-1.0 -lpango-1.0 -lgobject-2.0 -lgmodule-2.0 -lglib-2.0 -lintl -L/usr/nekoware/lib -rpath /usr/nekoware/lib -lfreetype -lz -lgc g++ -Wall -W -Wpointer-arith -Wcast-align -Wsign-compare -Woverloaded-virtual -Wswitch -Wno-unused-parameter -g -O2 -Wa,-xgot -o inkscape --export-dynamic main.o -L/usr/nekoware/lib libinkpre.a application/libinkapp.a ui/dialog/libuidialog.a dialogs/libspdialogs.a dom/libdom.a jabber_whiteboard/libjabber_whiteboard.a trace/libtrace.a svg/libspsvg.a widgets/libspwidgets.a display/libspdisplay.a helper/libspchelp.a libcroco/libcroco.a libnrtype/libnrtype.a libnr/libnr.a libavoid/libavoid.a livarot/libvarot.a ui/view/libuiview.a ui/libui.a ui/widget/libuiwidget.a extension/libextension.a extension/implementation/libimplementation.a extension/internal/libinternal.a extension/script/libscript.a xml/libspxml.a util/libinkutil.a io/libio.a inkjar/libinkjar.a libinkpost.a debug/libinkdebug.a -L/usr/nekoware//lib /usr/nekoware/lib/libgtkmm-2.4.so -L/usr/lib32 /usr/nekoware/lib/libgdkmm-2.4.so /usr/nekoware/lib/libatkmm-1.6.so /usr/nekoware/lib/libpangomm-1.4.so /usr/nekoware/lib/libglibmm-2.4.so /usr/nekoware/lib/libgtk-x11-2.0.so /usr/nekoware/lib/libgdk-x11-2.0.so /usr/nekoware//lib/libatk-1.0.so /usr/nekoware/lib/libgdk_pixbuf-2.0.so /usr/nekoware/lib/libpangocairo-1.0.so /usr/nekoware/lib/libcairo.so -lpng12 /usr/nekoware/lib/libglitz.so /usr/nekoware/lib/libxslt.so /usr/nekoware/lib/libsigc-2.0.so -lpng /usr/nekoware/lib/libgnomevfs-2.so -L/usr/local/lib -L/usr/local2/lib /usr/nekoware/lib/libxml2.so -lssl -lcrypto -lgen /usr/nekoware/lib/libbonobo-2.so /usr/nekoware/lib/libgconf-2.so /usr/nekoware/lib/libbonobo-activation.so /usr/nekoware/lib/libORBitCosNaming-2.so -lfastm /usr/nekoware/lib/libORBit-2.so /usr/nekoware/lib/libpopt.so /usr/nekoware/lib/libgthread-2.0.so -lpthread /usr/nekoware/lib/libXft.so /usr/nekoware/lib/libXrender.so -lX11 /usr/nekoware/lib/libfontconfig.so /usr/nekoware/lib/libexpat.so /usr/nekoware/lib/libpangoft2-1.0.so /usr/nekoware/lib/libpango-1.0.so -lm /usr/nekoware/lib/libgobject-2.0.so /usr/nekoware/lib/libgmodule-2.0.so /usr/nekoware/lib/libglib-2.0.so /usr/nekoware/lib/libintl.so /usr/nekoware/lib/libiconv.so -lc /usr/nekoware/lib/libfreetype.so -lz /usr/nekoware/lib/libgc.so -ldl -Wl,-rpath -Wl,/usr/nekoware/lib:/usr/nekoware//lib ld32: WARNING 84 : dom/libdom.a is not used for resolving any symbol. ld32: WARNING 84 : jabber_whiteboard/libjabber_whiteboard.a is not used for resolving any symbol. ld32: WARNING 84 : inkjar/libinkjar.a is not used for resolving any symbol. ld32: WARNING 84 : /usr/nekoware/lib/libgtkmm-2.4.so is not used for resolving any symbol. ld32: WARNING 84 : /usr/nekoware/lib/libxslt.so is not used for resolving any symbol. ld32: WARNING 84 : /usr/lib32/libgen.so is not used for resolving any symbol. ld32: WARNING 84 : /usr/nekoware/lib/libXft.so is not used for resolving any symbol. ld32: WARNING 84 : /usr/lib32/libdl.so is not used for resolving any symbol. ld32: ERROR 33 : Unresolved text symbol "Gtk::Main::Main(int*, char***, bool)" -- 1st referenced by main.o. Use linker option -v to see when and which objects, archives and dsos are loaded. ld32: ERROR 33 : Unresolved text symbol "Gtk::Main::run()" -- 1st referenced by main.o. Use linker option -v to see when and which objects, archives and dsos are loaded. ld32: ERROR 33 : Unresolved text symbol "Gtk::Main::~Main()" -- 1st referenced by main.o. Use linker option -v to see when and which objects, archives and dsos are loaded. ld32: ERROR 33 : Unresolved text symbol "Glib::ustring::ustring()" -- 1st referenced by libinkpre.a(file.o). Use linker option -v to see when and which objects, archives and dsos are loaded. ld32: ERROR 33 : Unresolved text symbol "Glib::ustring::operator=(char const*)" -- 1st referenced by libinkpre.a(file.o). Use linker option -v to see when and which objects, archives and dsos are loaded. [many many lines skipped] ld32: ERROR 33 : Unresolved text symbol "virtual thunk to Gtk::Adjustment::~Adjustment()" -- 1st referenced by extension/libextension.a(parameter.o). Use linker option -v to see when and which objects, archives and dsos are loaded. ld32: INFO 152: Output file removed because of error. collect2: ld returned 2 exit status gmake[2]: *** [inkscape] Error 1 gmake[2]: Leaving directory `/usr/people/voidfoo/devel/inkscape/inkscape_gcc/build_gcc/src' gmake[1]: *** [all-recursive] Error 1 gmake[1]: Leaving directory `/usr/people/voidfoo/devel/inkscape/inkscape_gcc/build_gcc' gmake: *** [all] Error 2 --- end pasting error msg --- Thank you very much! -- Jeremy Meng ---------------------------------------------------------------------------

4 5

NEW: Selected style indicator, statusbar redesign
by bulia byak 12 Nov '05

12 Nov '05

A new control in the left end of the statusbar lets you quickly view and change the fill and stroke of the selected object(s). * The two indicators, labelled "F:" (top) and "S:" (bottom), display fill and stroke of the selected object(s) correspondingly. * Each indicator can display either a color/opacity swatch (the opacity shown here is the fill opacity or stroke opacity, not the master opacity) or a label specifying "N/A" (nothing selected), "None" (no fill/stroke), "Unset" (unset fill/stroke), "L Gradient", "R Gradient", "Pattern" (corresponding fill/stroke types), or "Different" (selected objects have different stroke types). * Additionally, each indicator may be accompanied by one of two flags, "M" ("multiple", meaning there are two or more objects all with the _same_ fill/stroke) or "A" ("averaged", meaning there are two or more objects with flat color fill/stroke whose colors are averaged). * Just as the Fill&Stroke dialog, the new control correctly displays the fill/stroke of the selected gradient handle or selected text fragment in the corresponding tools. * Left-click on an indicator opens or activates the Fill&Stroke dialog with the corresponding tab selected. * Right-click on an indicator opens a popup menu with the following commands: ** "Remove fill/stroke": Removes fill or stroke from the selected objects. ** "Last set color": Applies to the selected objects the fill/stroke color that was last applied to anything. ** "Last selected color": Applies to the selected objects the fill/stroke color that was last displayed in this indicator. (To easily copy fill/stroke color between objects: select source, select destination, apply "last selected color".) ** "Unset fill/stroke": Unsets fill or stroke from selected objects (useful for clone originals). ** "Edit fill/stroke...": Opens or activates the Fill&Stroke dialog with the corresponding tab selected. * Middle-click on an indicator removes fill/stroke from selected objects; if it is already removed (i.e. if the indicator displays "None"), it does the same as the "Last set color" command from the popup menu. (In other words, it toggles between "None" and the last set color.) The zoom field and the cursor coordinates indicator have been rearranged for compactness and moved to the right end of the statusbar. == Implementation notes: == * There's no fixed pixel sizes for fonts used anywhere. Instead, the control uses a font size 70% of the current gtk default. I hope this works for most people, but I know there will be those who will find this font size too small, and others for whom it will be too large and pushing the vertical size of the statusbar. I want to hear your feedback on this. * The new control is itself done in gtkmm, yet it uses an old GTK color preview widget, and fits nicely into the old GTK desktop widget. The Glibmm::wrap() and gobj() are all you need to combine GTK and gtkmm stuff freely. So, I may be missing something, but I really don't understand why you gtkmm guys do this --new-gui thing instead of just converting the existing desktop widget-by-widget. * Despite lots of functionality, the new control is quite small in the code. That's because it reuses all the architectural concepts I've been slowly pushing in the last two years: desktop's current color, set and query style signals. With that in place, coding this entire control took just a couple days. It should also be easy to change or extend it. * TODO: I also plan to add a master opacity spinbutton to this control. -- bulia byak Inkscape. Draw Freely. http://www.inkscape.org

6 14

Re: [Inkscape-devel] Port to SGI IRIX 6.5
by Jeremy Y. Meng 11 Nov '05

11 Nov '05

On Sat, Nov 12, 2005 at 04:19:24AM +1100, Peter Moulder wrote: > >> >> +++ inkscape_irix/src/libavoid/makepath.cpp 2005-11-09 14:19:52.346046320 -0800 > >I'll add #include <algorithm> and `using' statements for make_heap, >push_heap, pop_heap. > No, please do not #include <algorithm>. It was my mistake to introduce it. std::-prefixing *_heap() would be enough. -- Jeremy Meng ---------------------------------------------------------------------------

2 2

0.43pre3: Everything is perfect
by Ted Gould 11 Nov '05

11 Nov '05

And, if it's not, now is a good time to mention it. http://sourceforge.net/project/showfiles.php?group_id=93438&package_id=9911… MD5 Sums: 3514904d0044f3670c5977d232f5ab87 inkscape-0.42+0.43pre3.tar.bz2 72fb0075f2465f990d36c08ee9fd29ad inkscape-0.42+0.43pre3.tar.bz2.sig d6140b948cc918773a4b414bdbdaf041 inkscape-0.42+0.43pre3.tar.gz 0d0d75c4f02eb01e71645c0f20f3ba03 inkscape-0.42+0.43pre3.tar.gz.sig --Ted

6 7

CVS
by Joshua A. Andler 11 Nov '05

11 Nov '05

Uhm... at least three hours now SF devel CVS has been borked. This is one of the longest outages I've seen and was wondering if we should file a trouble ticked with SF or what. -Josh

2 1