December 2007 - Inkscape Devel

the number of NEW bugs in the Tracker is now < 1000
by Ryan Lerch 17 Dec '07

17 Dec '07

Good news everyone, Your friendly Inkscape Bug Team has been busy triaging bugs over the weekend, and the total number of Bugs with a "NEW" status is now under 1000 (about 57% of total bugs.) When Triaging on Launchpad started a couple of weeks ago, this figure was up near 1900 (nearly 100% of all bugs) cheers, ryanlerch

2 1

reinterpret_cast unreliable on windows?
by bulia byak 17 Dec '07

17 Dec '07

I have fixed this windows-only bug, but I cannot fully understand how: https://bugs.launchpad.net/inkscape/+bug/169011 The problem was in this line: g_snprintf( b, len, "%s:%s;", key, css.gcharp() ); and gcharp is defined as: gchar const *gcharp() const { return reinterpret_cast<gchar const *>(ostr.str().c_str()); } The problem is that sometimes, rarely on some Windows systems and more frequently on others, gcharp() returned empty string for non-empty ostr. I replaced it with simply: g_snprintf( b, len, "%s:%s;", key, css.str().c_str() ); and the problem went away. Can anyone explain how this could happen? Is reinterpret_cast unreliable? Maybe we should eliminate it everywhere? -- bulia byak Inkscape. Draw Freely. http://www.inkscape.org

2 1

lxml + numpy (on OS X PPC but not only)
by jiho 16 Dec '07

16 Dec '07

Hi everyone, In order to provide out of the box python functionality on OS X for the upcoming 0.46 release, we'll bundle the python packages Inkscape uses in all flavors (i.e. ppc and intel, for every version of python). I would need someone who has these packages installed to send them to me: - lxml and numpy ppc with python 2.4.* - lxml and numpy ppc with python 2.5.* These two versions of python are *not* the one shipped with Mac OS X Panther or Tiger. They are typically obtained via MacPorts, Fink or the binary distribution of Python for Mac. MacPorts versions are in / opt/local/Framework/Python/.../lib/site-packages. Versions for the binary distribution of python are probably somewhere in /usr/local or in /Library/Frameworks/Python.framework/2.4/. I am looking for two directories in site-packages, unsurprisingly named lxml and numpy. If no-one has them already it would be really nice if someone could compile them for me on a PPC machine. This is really simple: - download lxml and numpy (from their respective homes, google is your friend ;)) - decompress them (usually already done for you) - in terminal sudo python setup.py install - go fetch the directories where they where installed (check the latest terminal messages), zip them and contact me. Thanks in advance. PS for the more knowledgeable around here: I am acting under the assumption that packages such as lxml and numpy are incompatible between point versions of Python (2.3 vs 2.4 vs 2.5), which is what I read somewhere. If someone has indication that this is not true, it would be very nice to point my mistake: things would be simpler and Inkscape won't grow too fat if we can just ship one version of the packages (the sum of all pyton packages will probably be around 70Mb, compared to the 80Mb of Inkscape itself, it seems a bit overkill...). JiHO --- http://jo.irisson.free.fr/

2 2

"make check" failures
by Ted Gould 16 Dec '07

16 Dec '07

Table of Contents: 1) Attributes -- looks like 3d-box tool 2) Style -- may be a bad test? 3) Bezier issue -- no clue 4) Rotate inversion -- no clue I'm not sure on all of these. I can of course start to look into them, but if you know you've modified these areas in the last year or so please look at the tests. Thanks, Ted --- 1 - Attribute test --- In AttributesTest::testAttributes: ./../../src/attributes-test.h:448: Error: Test failed: "azimuth" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:448: Error: Test failed: "elevation" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:448: Error: Test failed: "in" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:448: Error: Test failed: "limitingConeAngle" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:448: Error: Test failed: "pointsAtX" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:448: Error: Test failed: "pointsAtY" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:448: Error: Test failed: "pointsAtZ" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:448: Error: Test failed: "result" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:448: Error: Test failed: "z" ./../../src/attributes-test.h:448: Error: Expected (recognized == all_attrs[i].supported), found (true != false) ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 11 {inkscape:path-effect} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 53 {inkscape:snap-bbox} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 54 {inkscape:snap-nodes} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 55 {inkscape:snap-guide} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 56 {inkscape:snap-center} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 57 {inkscape:snap-intersection-grid-guide} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 58 {inkscape:snap-intersection-line-segments} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 61 {inkscape:bbox-paths} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 62 {inkscape:bbox-nodes} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 74 {inkscape:original-d} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 80 {inkscape:3dbox} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 81 {inkscape:box3dcornerA} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 82 {inkscape:box3dcornerB} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 83 {inkscape:box3dcornerC} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 84 {inkscape:perspective} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 85 {inkscape:box3dface} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 165 {flood-color} not handled" ./../../src/attributes-test.h:491: Warning: "Attribute string with enum 166 {flood-opacity} not handled" ./../../src/attributes-test.h:496: Error: Assertion failed: !found ./../../src/attributes-test.h:501: Error: Test failed: "Enum round-trip through string {flood-color} failed." ./../../src/attributes-test.h:501: Error: Expected (index == postLookup), found (165 != 257) ./../../src/attributes-test.h:501: Error: Test failed: "Enum round-trip through string {flood-opacity} failed." ./../../src/attributes-test.h:501: Error: Expected (index == postLookup), found (166 != 258) --- 2 - Style --- Testing style... sp_style_new, sp_style_write_string...Assertion `streq(str0_all, str0_all_exp)' failed sp_style_merge_from_style_string(whitespace, ifset...OK sp_style_merge_from_style_string(whitespace, always...Assertion `streq(str0_all, str0_all_exp)' failed sp_style_merge_from_style_string("fill:", ifset)...OK sp_style_merge_from_style_string("font-family:", always)...Assertion `streq(str0_all, str0_all_exp)' failed sp_style_merge_from_style_string("fill: ", ifset)...OK sp_style_merge_from_style_string("font-family: ", always)...Assertion `streq(str0_all, str0_all_exp)' failed sp_style_merge_from_style_string(":none", ifset)...OK invalid color setting: #4321...Assertion `streq(str0_set, "color:#123")' failed invalid color setting: currentColor...Assertion `streq(str0_set, "color:#123")' failed invalid color setting: #87654321...Assertion `streq(str0_set, "color:#123")' failed invalid color setting: #42...Assertion `streq(str0_set, "color:#123")' failed invalid color setting: aquam...Assertion `streq(str0_set, "color:#123")' failed sp_style_merge_from_style_string(default): ifset...Assertion `streq(ifset_str, str0_all_exp)' failed sp_style_merge_from_style_string(default): always...Assertion `streq(ifset_str, str0_all_exp)' failed sp_style_merge_from_style_string...Assertion `streq(str0_all, str0_all_exp)' failed SP_SCALE24_MUL...OK sp_style_merge_from_dying_parent: opacity...OK style: FAILED (6/18 tests passed) --- 3 - Bezier issue --- bezier_pt(1)...Assertion `bezier_pt(1, a, ti) == si * a[0] + ti * a[1]' failed sp_bezier_fit_cubic...Warning: got different control points from previously-coded (diffs=0x3c). Previous: (5, -3) (7.57134, -0.423509) (4.77929, 1.22426) (3, 3) Found: (5, -3) (7.32396, -0.46791) (5.06163, 1.32841) (3, 3) TODO: Update this test case for revised right_tangent implementation. --- 4 - Rotate issue --- inverse...Assertion `rot234.inverse() == NR::rotate(-.234)' failed

4 5

Grids
by Maximilian Albert 13 Dec '07

13 Dec '07

Hi all! Diederik, thanks a lot for enabling the axonometric grid in SVN! It's great to play around with it. Just one small thing I noticed: It doesn't honour the "snapping enabled" option in the Document Properties dialog. As I was giving it a try I was for a short moment a bit confused that the newly created grid (which had the checkbock "visible" ticked) didn't appear on the canvas - until I remembered to press '#' to make all grids visible. I still believe that users unfamiliar with the way it works may be quite confused by this behaviour. Any suggestions how to make the difference between enabling grids globally and individually more obvious to users? Would it be sufficient to add a caveat to the the tooltip in Document Properties? Max P.S.: Just noticed one other thing. When more than one grid is present then pressing '#' always switches to the tab of the first grid in Document Properties. This is a bit annoying when you actually want to work with the second grid, for example. Is there an easy way to change this?

9 17

The 3D box tool is dead. Long live the 3D box tool!
by Maximilian Albert 13 Dec '07

13 Dec '07

Hi all, I don't know how many times I announced this, but it has finally happened: As of rev #16681 the new, completely reworked(TM) 3D box tool code is now in SVN, hopefully still in time for 0.46. Those of you who played around with the tool before may have noticed that although it was certainly usable, there were a couple of issues; most notably, undo/redo didn't work reliably, some box configurations in degenerate perspectives were not really stable, etc. But most of all, development was getting more and more of a pain, making both debugging and adding new features tremendously difficult. All this was due to the internal architectural design of the tool at the time. Since I didn't see a way to easily sort this out by continuing the chosen path (and even if I had, it would only have led the tool even further into a dead end as far as extensibility and applicability to other 3D related tasks are concerned), I decided that it would be better to rewrite a lot of it more or less from scratch. I apologize that this has taken me so long, but I'm studying for my finals right now (or rather should be :-/ ), which has been a major distraction for completion. Anyway, the bulk is done now, and I'd like to give you a very quick overview of the changes. As far as the "ordinary" user is concerned, the overall behaviour of the tool is roughly the same as before, except that it is more stable and reliable. :) (In particular, there haven't been any great new features added yet). The most notable visible changes are the following: - Boxes now come equipped with a default depth, i.e. simply dragging on the canvas creates a full-featured box. This hides the allegedly hard-to-discover Shift+drag feature from the user and makes it easy for newcomers to find out what the tool actually does. - Boxes sharing the same perspective move together when a vanishing point is dragged, as if they form part of the same 3D scene that is simply viewed from a different angle (before, they all moved independently of each other). This makes it easy to create entire scenes and adjust them as a whole during drawing. If you want boxes to move independently of each other, you can split perspectives up by shift-dragging a vanishing point. For a full list of this and other features of the tool, please wait until I have updated the Release Notes (a copy will be posted to this list). So much for the visible part of the tool. For those who are interested here comes a more in-depth description of the internal changes: 1) *All* 3D box related structures (namely, the box itself, its sides, and also perspectives) are now handled as SPObjects, which fixes the annoying issues with undo/redo. A notable consequence is that the SVG representation of boxes is now slightly different than it was before, which probably renders old files invalid or even unusable. This shouldn't be a great issue, though, since the tool has not officially been enabled yet (which turns out to be a good decision). I'm going to give the current representation a final revise in the next one or two days, but I don't expect it to change any more. 2) Perspectives are now added to the document's defs so that they can in principle be reused by other objects. This doesn't have an immediate effect on the tool itself but should make applications to other 3D related situations easier. 3) Before the refactoring, all constructions were made geometrically by computing connecting lines, points of intersection, etc. The way it works now is that we use the information given by the three vanishing points to compute a mapping from 3D space to the canvas which results in the specified perspective. By computing inverse images of canvas points, this allows us to determine "real" 3D coordinates of boxes, to model the concept of lengths (at least to a certain extent), to compare coordinates of boxes, etc. Also, degenerate perspectives can be handled in a numerically stable way because not the boxes themselves are degenerate but only their _images_ on the canvas. In fact, this description is a bit of an oversimplification. In order to compute the mentioned perspective mapping, it is necessary to (at least implicitly) specify a _fourth_ point on the canvas, which can be thought of as the origin of the coordinate system (whose three axes point towards the vanishing points). This fourth point is not currently represented on screen. A while ago in one of our discussions Bulia suggested that it should be hidden from the user because its meaning might be unintuitive. Personally, I'm not entirely convinced and I'm tempted to leave it as an option to the user whether it should be drawn on the screen or not (it can also be quite useful for adjusting 3D scenes). Any opinions from other people? At present you can visualize this point by pressing 'O' in the 3D box tool (you may need to draw a box first). This draws a small pink point on the canvas. Note that I added this shortcut for debugging purposes and it will be removed later on. Think of this point as the "anchor" of the coordinate system - it is the only point that always remains fixed when vanishing points are dragged. Well, that is pretty much all I can think of at the moment. Please ask if you are missing anything or if something remained unclear. In principle, everything should more or less work now. There are a few known issues I'm going to sort out in the next days, including (but probably not limited to) the following: - I had to temporarily disable the spin buttons in the toolbar after introducing an undo-related crash yesterday (Inkscape chokes with the message "assertion failed: (doc->priv->sensitive)"). I will debug this and make them work again very soon. In the meantime, the states of vanishing points can be toggled using the shortcuts "X", "Y", and "Z". In addition, the angles of infinite perspective lines can be rotated using the various types of brackets - namely [], (), {} (where Alt+dragging results in a very small rotation angle). - Although z-orders should be set reliably now, there may be a few degenerate cases where they are still wrong (I think I know which these are). If you discover such a case, you can send me a corresponding *.svg file if you wish. - Transformations of boxes work, but if a single box is transformed this also transforms all other boxes sharing the same perspective. Bulia said that instead this box needs to be unlinked. Will fix this soon. - Stroke, patterns, and gradients are not yet adapted during transformations. I'll have to see how to do this because transformations are implemented differently in 3D boxes than in the other tools. Any general pitfalls I need to know or keep in mind regarding this? - Constrained dragging of handles (by using Ctrl) somehow sets the box off a certain amount from the intended position. Don't know precisely what causes this but I hope it won't be hard to fix. - flipping boxes doesn't update its z-orders - ... whatever you may discover ... As always, if there are any questions, comments, suggestions, you name it - please don't hesitate to tell me. And now: Happy testing and enjoy. :=) Cheers, Max P.S.:There is still a number of debugging messages present in the code (most of them commented out, but maybe I forgot some). I will of course get rid of them soon, but currently they are helping me in resolving the remaining issues, so I left them in.

5 8

compile error on windows
by bulia byak 13 Dec '07

13 Dec '07

trying to compile with bob's latest mingw: In file included from src/transf_mat_3x4.cpp:17: src/syseq.h: In function 'void SysEq::fill_random(double (*)[T], double, double)': src/syseq.h:74: error: there are no arguments to 'random' that depend on a template parameter, so a declaration of 'random' must be available src/syseq.h:74: error: (if you use '-fpermissive', G++ will accept your code, but allowing the use of an undeclared name is deprecated) -- bulia byak Inkscape. Draw Freely. http://www.inkscape.org

3 3

Error: Unknown verb name
by Maximilian Albert 13 Dec '07

13 Dec '07

Hi, for a while now I have been getting the following error messages during startup if Inkscape: ** (inkscape:6536): WARNING **: Unknown verb name: EditCloneOriginal Unable to find: EditCloneOriginal ** (inkscape:6536): WARNING **: Unknown verb name: EditCloneOriginal Unable to find: HelpKeys ** (inkscape:6536): WARNING **: Unknown verb name: HelpKeys Unable to find: org.inkscape.help.keys Even "make clean" didn't help. Any ideas where they come from? I can't imagine that everyone is seeing this because no one else has complained so far. Thanks! Max

2 3

Process for patching 2geom
by John Faith 13 Dec '07

13 Dec '07

Hello, A while ago I submitted a patch which helps OS X build to the 2geom SourceForge tracker assuming that the patch would be added to 2geom, then 2geom would be updated in inkscape source. Is this correct or should I submit a patch to the new launchpad system? It's been about a month and nothing has happened with the patch, so I guess I must have done something wrong. Thanks!, John

3 2

Inkscape overcomes Xara in Google Trends
by bulia byak 12 Dec '07

12 Dec '07

http://www.google.com/trends?q=inkscape%2Cxara Note that "xara" does not always refer to the vector editor, so in reality it's even better for us. Now, to support the momentum, we must release a smashing 0.46 soon enough, and make a lot of noise about it. It really deserves it. (Note by the way that while we are ahead in user interest, we are nowhere to be seen on the lower graph which reflects mentions in the news. That's not fair.) GSoCers and all others, please let's get to work one more time and fix all the loose ends in your projects. They need not be perfect but the most glaring user-visible problems must be ironed out for Inkscape to reach a releaseable state. You know who you are and what you need to do :) If you think that your project is already good enough for a release, please drop a note here, otherwise let us know of your current state, what remains to do, and when you are planning to finalize your work. And, let's make bugfixing a priority right now, even without waiting for a freeze. We have a scary number of open bugs in the tracker. Yet we now have so many developers that if each one closes just one bug a day, we'll make the tracker shrink pretty fast. Note that closing a bug need not necessarily involve coding; you can just take some old bug, test it and close as fixed or unreproduceable. Any help from non-developers is of course very welcome too. Let's all test, report, comment, add info, code, test again, and close as many bugs as possible. -- bulia byak Inkscape. Draw Freely. http://www.inkscape.org

19 38