- Inkscape Devel - Inkscape Mailman

Fwd: Set new coordinates for points in thePiecewise<D2<SBasis> > array
by Anton Moiseev 26 Jan '10

26 Jan '10

Hello, > I haven't read your mail in much detail, but generally it is a very bad idea to directly modify nodes of a path. > Could you exp > Actually, the effect is live version of "perspective" extension. The input is rectangle with modifiable by user nodes (which is originally generated from the object bounding box), the result - the object would be inscribed into modified rectangle, so it would look like in perspective. I am not aware if this problem has a solution based on some kind of generic coordinate transform formula, which could be applied to the whole Piecewise object, so I use my own solution of the problem based on straight-forward linear geometry, which calculates new position for each object's point and then tries to apply this value to the original point - as if the user used "node editor" tool to move points to the places where he wants to. I can try to describe the algorithm in more details if that makes sense. I know there is experimental perspective_path live effect which also works with perspective, but when I tried it, it appeared that it works in some different way - it does not use bounding rectangle to setup perspective, but relies on object resize/movements instead, so this is not what I need exactly. I also looked at the perspective.py code, but it seems that python extensions work in a bit different way than live effects, so I could not neither reuse it as is, nor understand its formula basing on source code without comments. thank's Anton

2 2

Set new coordinates for points in the Piecewise<D2<SBasis> > array
by Anton Moiseev 26 Jan '10

26 Jan '10

Hello, I'm trying to write a new live effect which would move original path points to new locations. I have taken lpe-envelope.cpp for the base and write my code inside Geom::Piecewise<Geom::D2<Geom::SBasis> > LPEPerspective::doEffect_pwd2 (Geom::Piecewise<Geom::D2<Geom::SBasis> > const & pwd2_in) so, I receive Piecewise<D2<SBasis> > at the input and want to return generated object Piecewise<D2<SBasis> > at the output, which would contain all orinal pwd2_in points, but with new coordinates. Generally I did not have problems with my algorithm and I have an array of Point objects, but now when I want to put them back to Piecewise<D2<SBasis> > array, the result does not have the required quality: It is hard to explain in more details, so would try to put some code I have. Piecewise<D2<SBasis> > output = pwd2_in; // this gives me a path with same number of nodes as the original one then I go though all points in the path and try to set new location for each point for (unsigned i = 0; i < pwd2_in.segs.size(); i++){ Point target_point = XXX; // calc new value for current point output[i][0][0][0] = target_point[0]; // set value for x output[i][1][0][0] = target_point[1]; // set value for y } return output; With code which works in this way the transformation converted any the object to invisible phantome (the selection showed its area with nothing inside, but when remove the selection it is not possible to select object again, also the "edit points" did not show any objects) - in short, the result was broken. Then I played a bit with printing values for pwd2_in path, understood that I must somehow play with SBasis object values (set values to SBasis[0][0] and SBasis[0][1]) and added the following to the code: for (unsigned i = 0; i < pwd2_in.segs.size(); i++){ Point target_point = XXX; // calc new value for current point output[i][0][0][0] = target_point[0]; // set value for x output[i][1][0][0] = target_point[1]; // set value for y if(i > 0) { // setup previous point target value output[i - 1][0][0][1] = target_point[0]; output[i - 1][1][0][1] = target_point[1]; } else { output[i][0][0][1] = target_point[0]; output[i][1][0][1] = target_point[1]; } } return output; That was much better - I was able to see the deformed objects which looked much like originals, but still had some issues: 1. Bended curves became straight lines 2. I have problems with objects consisting of multiple paths Actually, the 1st point only shows itself when I try to deform such object as symbol '3' (converted to path) - after deforming its shape looks like original, but consists of straight lines; though when I deform a simple triangle with one rib curved,the curve does not disappear, but did not test much though with this deffect. The 2nd problem for now is more important - it is visible after deforming symbol '4' - when it is converted to path, it consists of 2 unconnected path objects, and after my deforming, those 2 paths become connected at some point. This is definitely happens because I set incorrect values to SBasis[0][1] and actually I would prefer not to play with SBasis object by hands and to do my task in some other more reliable way, but for now I could not find anything relevant in the documentation. thank's

2 1

Call for booth volunteers - SCALE8x
by Jon Cruz 26 Jan '10

26 Jan '10

Hello, This is a general call to see if anyone might be able to volunteer some time to help man the Inkscape booth/area at SCALE8x next month in Los Angeles. If you might be able to lend some time, we have a few conference passes as an exhibitor. So if you can get there, we can get you in. :-) It's a great experience, and very low pressure. Don't feel intimidated; step on up.

1 0

Devlibs in Launchpad?
by Krzysztof Kosiński 25 Jan '10

25 Jan '10

Can I put the Windows devlibs in a Launchpad project? (not in lp:inkscape, but in a separate project like lp:inkscape-windows-devlibs) I have successfully compiled Inkscape on both Linux and Windows using Waf, but there are 2 or 3 minor changes I need to make to the Windows devlibs. Moving them to Launchpad would make them more accessible to all developers. Regards, Krzysztof

4 5

Re: [Inkscape-devel] Node tool committed to trunk
by Krzysztof Kosiński 25 Jan '10

25 Jan '10

W dniu 25 stycznia 2010 06:28 użytkownik ~suv <suv-sf@...58...> napisał: > Can I ask you two more general questions related to changes from your > node-tool merge? (I'm just curious and trying to better understand): > > a) date removed from about-dialog > I reverted this change locally because when testing different builds > e.g. with different patch versions from the bug tracker that don't > change the revision number - or re-building after external changes - I > thought the build date at least as useful as the revision number (during > development phases I wouldn't mind including the build time as well). > I'd like to understand the reason to reduce the 'About' info to the > revision number only, while trunk is still open for so many changes? The problem was that the date displayed in the dialog was not the actual date of the build, but the date of building the aboutbox.cpp file. Since that file doesn't change too often, it would not be accurate. I changed it so that the date was included in the version string that gets written to inkscape-version.cpp, but probably it got lost during the Bazaar transition. > b) src/ui/icon-names.h > When comparing available lists of icon names (wiki, src) I noticed that > your new 'node-transform' icon has no entry in the macro list for icon > names. Will the list in 'src/ui/icon-names.h' no longer be the most > up-to-date list of named Inkscape icons? I think that file wasn't a very good idea to begin with - changing the names will cause the define names to be out of sync with actual names, which is confusing, and changing any icon name or adding one will cause extensive rebuilds, because icon-names.h is included in many files. I have a different solution to the "list of all icons" problem that I'll share later. > p.s. btw - I am *very* pleased with the new transformation handles for > selected nodes: it is the first time ever I can savor the ability to > rotate nodes! Previously this action was accessible only by keyboard > shortcuts which completely fail with many localized keyboard layouts > ([,]) *and* - iiuc - bypass verbs and thus any custom keyboard shortcut > mapping ;( I'll consider adding a verb or GtkAction for each keyboard action in the node tool. This problem of non-configurable keybindings is not limited to the node tool, by the way, though probably it's the most noticeable in the node tool due to the large number of shortcuts it has. Regards, Krzysztof

1 0

Re: [Inkscape-devel] Waf build branch
by Krzysztof Kosiński 25 Jan '10

25 Jan '10

W dniu 25 stycznia 2010 04:42 użytkownik the Adib <theadib@...1439...> napisał: > Krzysztof, > > does it build parallel (aka make -j2)? > > Thx, Adib. Yes. If you don't give it any parameters it will use all available logical CPUs. I use it on an i7 (8 cores) and in a Windows VM with 2 cores without any issues. I'll put detailed instructions on how to build on Windows using devlibs and the waf-build branch on the wiki, hopefully today or tommorow. Regards, Krzysztof

1 0

Waf build branch
by Krzysztof Kosiński 24 Jan '10

24 Jan '10

I pushed the Waf build system branch as: lp:~tweenk/inkscape/waf-build I would enjoy if someone tested it on Mac OS X and Windows, and reported how it doesn't work. :) The commands to build Inkscape are (in the top level source directory): ./waf configure ./waf build Things that should work on Linux: - building - updating translation templates (via ./waf i18n) - installation Regards, Krzysztof

4 9

Re: [Inkscape-devel] Help regarding python extensions
by Jonatã Bolzan 21 Jan '10

21 Jan '10

On Thu, Jan 21, 2010 at 6:49 AM, Jonatã Bolzan <jbopen@...400...> wrote: > Well, if you say that... here I go :D > > I want the extension passing for all the nodes. To be more clear, here I > put an example of code from the extension "svg_and_media_zip_output.py": > > for node in self.document.xpath('//svg:image', namespaces=inkex.NSS): > > This line get every "image" object in the svg. In my case, I want to get > everything that is drawn in SVG. I want images, shapes, text... do you > understand? > > I know it's a bit silly question, but will solve some issues here. > > Thanks for the attention. > > On Wed, Jan 20, 2010 at 8:42 PM, Cédric Gémy <radar.map35@...8...> wrote: > >> Yep post here. I would also say that it would be great if we'd manage to >> update wiki pages for extensions. They seem quite outdated. I wanted to >> write some extensions few weeks ago (i already done some earlier) and >> found that many things have changed so that the actual doc is not >> enough. >> >> >> >> >

1 0

Help regarding python extensions
by Jonatã Bolzan 20 Jan '10

20 Jan '10

Hi there! I want to know if there are somebody here who know about python extensions for Inkscape and have enough patience to answer some questions :D . If so, please contact me at jbopen@...400... Thanks anyway for the attention.

3 2

Warning cleanup of hash-iness
by Jon Cruz 20 Jan '10

20 Jan '10

Well... it looks like Krzysztof found us a nice, tricky problem in regards to the whole hash_map/unordered_list mess. What he has been tripped up by is one of those ugly problems with no simple solutions. First, we have legacy code that is using a gcc extension. Not so good. Next the replacement has problems on many systems. So we don't have a tidy solution. General questions for any thing with warnings, etc. * Are the warnings critical? That is, do they indicate some immediate risk/failure? If so, then we should address them sooner * Do we need to be using it? That is, are there any alternatives to the code creating warnings? Things such as vector instead of list, map instead of hash_map, etc. * Can we encapsulate it? That is, do we need to leave things in .h files that other things include, or can they be moved inside specific files? * Is there a clean way to fix at least some of the warnings? For some, this may be as simple as using an explicit variable instead of a temporary. In this case there appears not to be * Is it cross-platform? Remember, we need to keep Inkscape compatible with multiple platforms, including MS Windows, Linux, OS X, BSD, Solaris, etc. I'll try to follow up on a few of these points.

4 16