April 2005 - Inkscape Devel - Inkscape Mailman

Refactoring in action
by Jon A. Cruz 22 Apr '05

22 Apr '05

Just thought I'd toss out this note "Refactoring in action" or "How I implemented an RFE in under an hour" This morning I changed things so that the CMYK color slider mode now has values that range form 0 through 100 instead of 0 through 255. In and of itself that was no big deal, and the ammount of code change needed was minimal. What is interesting, however, are the principles that came into play in doing it. ------------------------ I addressed RFE 994177 "CMYK 0 - 100 not 0 - 255". The first thing was to see if it was needed. Having worked with various media places and artist before, this was a simple one to see the utility of. Not earth-shattering or anything, but something that would make artist and designers working in print a little more comfortable. So, yes, it should be done at some point. Next question was how much effort would it take. Since I'm in the middle of addressing color issues in general, a lot was fresh in my mind. Then looking into the code I saw it was in very good shape, and prime for reuse. So the level of effort needed should be low. Now came the fun part. As I went into the code, it was pretty clean and easy to understand. More importantly, it already was setup to mainly separate the UI presentation from the underlying data model itself. In this case, the sliders would need to change depending on the 'mode'. But... the slider widgets were not played with directly, but instead were attached to instances of GtkAdjustment. It turns out that GtkAdjustment is pretty much a 'model' class to represent ranging data in an abstract way. So... my next step was to look at what needed changing. I first was going to set things up to change the sliders manually when the mode was switched. However, as I moved forward I realized that things were almost clean, and only needed to get set. I just added a 'range limit' member to track which limit the range was using. Now the really good news was that the functionality I needed to hack on was already moved out to a pair of common functions (get1 and set255). And instead of needing to change those to take a parameter, I just needed to make the code respect the limit already a member of the 'model' class being used (GtkAdjustment). I no longer needed to keep a parameter, and so didn't need ot pass it in or have different functions to do different values. I did rename those two, just to make things a little clearer on their use. But that was just a minor tweak to make things easier to read down the road. I did a little find/grep magic to check where things were used and clean them up, but I was done fairly early. Building and testing took up more of the time. Here is a quick summary of the files I changed: src/widgets/sp-color-scales.h 1.8 src/widgets/sp-color-scales.cpp 1.13 src/widgets/sp-color-slider.cpp 1.10 src/widgets/sp-color-wheel-selector.cpp 1.9 sp-color-scales.h: * renamed get1() to getScaled() * renamed set255() to setScaled() * added _rangeLimit (this actually can go away now) sp-color-scales.cpp: * set default range limit * use rangelimit in constructing adjustments * changed get1() and set255() to meet the GtkAdjustment limit instead of hardcoded values * set the range limit when switching modes sp-color-slider.cpp: * only the get1/set255 function renaming sp-color-wheel-selector.cpp: * changed _adjustmentChanged to meet the GtkAdjustment limit instead of a hardcoded value. ======================= Summary: Things went very fast because of a few things: * The code was fairly clean and easy to understand to begin with. * Common functionality was pulled out and centralized into common functions instead of being repeated in-line. (fix it in one place, and it was fixed everywhere) * There was a nice separation of UI and data model. Functional logic dealt with the data model, while the UI was just a slave to it. (in this case there was the RGBA color value and the GtkAdjustment models in play) Those combined to make it quite easy to change the UI without negatively impacting the functionality of Inkscape itself. The more we can allow for this type of refactoring, the better Inkscape can be. So I'd like to encourage everyone to try to keep principles like this in mind. Oh, and a big thanks also goes out to all the developers who were in that code before me.

3 2

Effects Tarball
by unknown＠example.com 22 Apr '05

22 Apr '05

I created a few more effects today. A path randomizer that randomly nudges path nodes and control points within a specified radius. A path segment straightener that straightens bezier path segments in two different ways. And a random tree generator. The tree generator uses a really simple little Turtle class that draws with SVG path data instead of pixels. Very basic but I think it should be enough to make an Inkscape effect out of just about any of those fun little fractal routines done in LOGO. A tarball of effects is available to try out: http://www.ekips.org/comp/inkscape/extending.php Aaron Spike

4 5

Aaron's effects screenshot
by Ted Gould 22 Apr '05

22 Apr '05

Hey all, Aaron made a screenshot of his effects, everyone should take a look. Very cool stuff! http://www.ekips.org/comp/inkscape/inx/InkexEffects.png --Ted

5 7

Re: [Inkscape-devel] screwy export
by bulia byak 21 Apr '05

21 Apr '05

On 4/21/05, John Taber <jtaber@...480...> wrote: > On Thursday 21 April 2005 07:17, bulia byak wrote: > > If by adjusting you mean scaling, just scale with Ctrl, or click the > > lock toggle in the selector controls. > okay but where is this lock or selector controls? I can't find it anywhere - > not under doc prefs, selecting. I think it should be under object properties > - that's the first place we looked for it. No. It's not a property of an object. It's a property of the tool which you use for scaling. Therefore, it's in the tool controls bar. It's the horizontal bar just above canvas. Use the button with the lock icon. > > > 2) clipping out what you want and pasting it into a new drawing - > > > then exporting it. > > There's no need to do this unless you want to hide all other objects > > except those you are exporting. Just select something and export > > consider that we use svg maps - a river runs the length of the drawing but we > are only clipping out an area that contains part of the river - that means > clipping the river object (assuming it's something like a bezier line) So do you want to hide all other objects when exporting the river or not? If not, just select the river and do export. If yes, you can do this from command line with the --export-id-only option (see man inkscape). > Okay, I understand what you are saying about the canvas and that we only need > to worry about the actual svg elements, but we want to take the above clipped > section, draw some stuff on it, resize, and export to a bitmap to put in a > report, where one bitmap export might be the whole clipped area, one might be > a zoomed in detail look. Does this make more sense of what the problem is ? I don't see what the problem is. Select an object and export it, this will clip your drawing by this object. If you want to export a detail, draw a transparent rect over it and export that rect. -- bulia byak Inkscape. Draw Freely. http://www.inkscape.org

1 0

Available export formats
by Lucas Vieites 21 Apr '05

21 Apr '05

Hi, I'm in need of knowing the available export or interchange formats for Inkscape. I normally use Inkscape to create graphics for the web, but now I created a design for a publicity flyer and the printer (the guys who are going to print the flyers) only work with CorelDraw 10. How can I export the svg to a format Corel understands. I've been googling a bit and I know newer versions of Corel import svg directly (at least version 12) but I'm stuck on this one. Can anybody help me? Thanks a lot in advance, -- Lucas Vieites Fariña Dept. Desarrollo <lucas@...212...> Asix Informática <http://www.asixinformatica.com/> Tel/Fax: +34 986 54 26 98

2 2

screwy export
by Jon Phillips 21 Apr '05

21 Apr '05

When I use export, it seems screwy with the DPI. I expect that if I change the DPI to 72, and then change the size of the exported image, the DPI should stay the same. It is really wacky and unlike other graphics apps for the DPI to change when I change the image size, and vice-versa. Actually, most apps, like gimp, etc have locks to constrain dpi and proportions (see attached screenshot). Anyhow, export use to work well for me to do exporting for web, but now is strange. Is this a problem for anyone else? Yes, I understand the logic that the size changes because there are then more pixels per inch, but I don't agree in how it acts. Once someone sets dpi in this, then the exported image should stay the same size and just have a higher DPI, but same size. Thoughts on this? Am I wrong here? Am I smoking something? Whats the deal?? Jon -- Jon Phillips USA PH 510.499.0894 jon@...235... http://www.rejon.org Inkscape (http://inkscape.org) Open Clip Art Library (www.openclipart.org) CVS Book (http://cvsbook.ucsd.edu) Scale Journal (http://scale.ucsd.edu)

3 7

Extensions Package
by Ted Gould 20 Apr '05

20 Apr '05

We've talked about before, and Aaron and I have recently talked about it, should we package extensions? I'd like put some ideas and questions forward so that people can rip them apart and we'll end up with something good :) The basic idea is that we may want to start shipping extensions packages somehow separately from the core Inkscape package. There are several reasons for this: -- Dependencies. The dependency checking in Inkscape will always be a toy when compared to dpkg and rpm. And, we don't want to require every package in Debian just to install Inkscape. -- Versioning. It might be more reasonable to have effects on a different release schedule if there were a larger collection of authors. Something more like OpenClipart's monthly release. -- Modularity. If we ever expect for people to be able to ship extensions for Inkscape from outside the project umbrella the tools to do so need to be good, and packing some of our own extensions externally would force use to make them reasonable. Reasons to just throw everything into the Inkscape tarball and live with it: -- Simple. Releases are a pain, more releases need to be justified. -- Complete Solution. Have you ever seen a public forum where someone posts a new program, and then 90% of the questions are "What theme did you use in the screenshots?" Multiple places to get extensions means that everyone won't have the same set. I think, that there is compelling reasons to start a new package for extensions, which brings the next points: how should it work? -- Granularity. Should each extension be it's own package? Should they be grouped by functionality (input/output/effect)? Should they be grouped by common dependencies (inkscape-extensions-{python,perl,pstoedit}) -- Versioning. I'm kinda thinking that we should just do <inkscape major>.<inkscape minor>.<extension release> That way extensions could be released independent of Inkscape, but still, you know what version of Inkscape they are tied to. Okay, I think that was verbose enough for now, if you made it this far I'm impressed. --Ted

2 1

Supporting <style> elements
by Bob Jamison 20 Apr '05

20 Apr '05

Hi, all, Last night, user <giordano> came onto the channel and showed an SVG file exported from Visio, which did not render properly on Inkscape. Turns out that the file has a separation of style specification into <style> and class="" references like this: <style type="text/css"> <![CDATA[ .st1 {fill:none;stroke:none;stroke-linecap:round;stroke-linejoin:round;stroke-width:0.72} .st2 {fill:#ffffff;stroke:#000000;stroke-linecap:round;stroke-linejoin:round;stroke-width:0.72} ...etc.... and later referenced like: <rect v:rectContext="foreign" x="0" y="892.83" width="2327.33" height="807.957" class="st1"/> I know that this is exactly the missing CSS functionality that we have on the roadmap, but is it really that hard to implement? I made a quickie Java class that loads up the css definitions into a HashMap, and later when the class="" tags use the refs, replaces the class="" with style="". So that line would look like this now: <rect v:rectContext="foreign" x="0" y="892.83" width="2327.33" height="807.957" style="fill:none;stroke:none;stroke-linecap:round;stroke-linejoin:round;stroke-width:0.72"/> Could we not, in the interim before we integrate full libcroco functionality, parse the <style type="text/css"> element into a std::map<std::string, std::string> and do the same substitution at load-time? I attached the Java source to show how simple this method would be. Bob import java.io.File; import java.util.StringTokenizer; import java.util.HashMap; import java.util.Iterator; import javax.xml.parsers.DocumentBuilder; import javax.xml.parsers.DocumentBuilderFactory; import javax.xml.parsers.ParserConfigurationException; import javax.xml.transform.dom.DOMSource; import javax.xml.transform.stream.StreamResult; import javax.xml.transform.Result; import javax.xml.transform.Source; import javax.xml.transform.Transformer; import javax.xml.transform.TransformerFactory; import javax.xml.transform.TransformerConfigurationException; import org.w3c.dom.Document; import org.w3c.dom.Element; import org.w3c.dom.Node; import org.w3c.dom.NodeList; import org.xml.sax.SAXException; class VisioFilter { HashMap<String, String> styles; void error(String msg) { System.out.println("VisioFilter:" + msg); } String getElementText(Element elem) { StringBuffer buf = new StringBuffer(); NodeList nl = elem.getChildNodes(); for (int i=0 ; i<nl.getLength() ; i++) { Node n = nl.item(i); int type = n.getNodeType(); if (type == Node.TEXT_NODE || type == Node.CDATA_SECTION_NODE) { buf.append(n.getNodeValue()); } } return buf.toString(); } boolean processStyleElement(String str) { StringTokenizer bigSt = new StringTokenizer(str, "\n\r"); while (bigSt.hasMoreTokens()) { String line = bigSt.nextToken(); //System.out.println("line:"+line); StringTokenizer st = new StringTokenizer(line, "{}"); if (!st.hasMoreTokens()) continue; String name = st.nextToken(); name = name.trim(); if (!name.startsWith(".")) continue; name=name.substring(1); if (!st.hasMoreTokens()) continue; String style = st.nextToken(); styles.put(name, style); } /* for (Iterator<String> iter = styles.keySet().iterator() ; iter.hasNext() ; ) { String key = iter.next(); System.out.println("name:'"+key + "' style:'"+styles.get(key)+"'"); } */ return true; } void substituteClassAttribute(Element elem) { String className = elem.getAttribute("class"); if (className!=null && className.length()>0) { //System.out.println("found class:"+className); String style = styles.get(className); if (style!=null) { elem.removeAttribute("class"); elem.setAttribute("style", style); //System.out.println("replaced with:"+style); } } NodeList nl = elem.getChildNodes(); for (int i=0 ; i< nl.getLength() ; i++) { Node n = nl.item(i); if (n.getNodeType() != Node.ELEMENT_NODE) continue; Element childElem = (Element)n; substituteClassAttribute(childElem); } } public boolean processFile(String inFile, String outFile) { System.out.println("############### PARSING"); DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance(); DocumentBuilder db = null; try { db = dbf.newDocumentBuilder(); } catch (ParserConfigurationException pce) { error("processFile:" + pce); return false; } File inf = new File(inFile); Document doc = null; try { doc = db.parse(inf); } catch (Exception se)//saxexception + ioexception { error("processFile:" + se); return false; } NodeList nl = doc.getElementsByTagName("style"); if (nl.getLength() < 1) { error("no style elements"); return false; } for (int i=0 ; i<nl.getLength() ; i++) { Node n = nl.item(i); if (n.getNodeType() != Node.ELEMENT_NODE) continue; Element elem = (Element)n; String type = elem.getAttribute("type"); if (type == null || !"text/css".equals(type)) continue; String val = getElementText(elem); processStyleElement(val); } System.out.println("############### PROCESSING"); substituteClassAttribute(doc.getDocumentElement()); System.out.println("############### WRITING"); File outf = new File(outFile); TransformerFactory tf = TransformerFactory.newInstance(); Source source = new DOMSource(doc); Result result = new StreamResult(outf); Transformer trans; try { trans = tf.newTransformer(); trans.transform(source, result); } catch (Exception tce) { error("processFile:" + tce); return false; } return true; } public VisioFilter() { styles = new HashMap<String, String>(); } public static void main(String argv[]) { if (argv.length!=2) { System.out.println("usage: VisioFilter infile outfile"); return; } String inFile = argv[0]; String outFile = argv[1]; VisioFilter vf = new VisioFilter(); vf.processFile(inFile, outFile); } }//end class

1 1

Re: [Inkscape-devel] Effects Tarball
by Alexandre Prokoudine 20 Apr '05

20 Apr '05

On 4/20/05, bulia byak <buliabyak@...400...> wrote: > On 4/20/05, Alexandre Prokoudine <alexandre.prokoudine@...400... > wrote: > > btw, is it possible to add a "_" to "Effects" (e.g. "E_ffects") so > > that this menu was selectable from keyboard at once? > > No because "f" is already taken for _File. And "e" for _Edit. You can > only use "c" or "s". Well :) I don't insist on _f :) It's just it should be there, no matter what exactly unique character is used. Alexandre

1 0

More warnings with gcc4.0
by Bob Jamison 20 Apr '05

20 Apr '05

Playing more with the current pre-release of gcc4.0, I found some more warnings due to its greater strictness. Here are the warnings I got with CVS today. 'make clean ; make > /dev/null' is great for getting this listing. People might want to look at their files. I can clean up some of these tomorrow. Actually, this is not very many! Bob dropper-context.cpp: In function 'void sp_dropper_c32_color_copy_hex(guint32)': dropper-context.cpp:216: warning: unused variable 'pick' file.cpp: In function 'void file_import(SPDocument*, const gchar*, Inkscape::Extension::Extension*)': file.cpp:752: warning: unused variable 'repr' select-context.cpp: In function 'gint sp_select_context_item_handler(SPEventContext*, SPItem*, GdkEvent*)': select-context.cpp:276: warning: unused variable 'selection' sp-item.cpp: In function 'SPItem* sp_item_first_item_child(SPObject*)': sp-item.cpp:1277: warning: control reaches end of non-void function sp-pattern.cpp: In function 'void sp_pat_fill(SPPainter*, NRPixBlock*)': sp-pattern.cpp:909: warning: operation on 'px_x' may be undefined sp-pattern.cpp:940: warning: operation on 'px_x' may be undefined dialogs/stroke-style.cpp: In function 'void sp_stroke_style_any_toggled(GtkToggleButton*, SPWidget*)': dialogs/stroke-style.cpp:1704: warning: unused variable 'items' display/canvas-arena.cpp: In function 'void sp_canvas_arena_class_init(SPCanvasArenaClass*)': display/canvas-arena.cpp:91: warning: invalid access to non-static data member 'SPCanvasArenaClass::arena_event' of NULL object display/canvas-arena.cpp:91: warning: (perhaps the 'offsetof' macro was used incorrectly) display/sp-canvas.cpp: In function 'void sp_canvas_item_class_init(SPCanvasItemClass*)': display/sp-canvas.cpp:110: warning: invalid access to non-static data member 'SPCanvasItemClass::event' of NULL object display/sp-canvas.cpp:110: warning: (perhaps the 'offsetof' macro was used incorrectly) libnrtype/Layout-TNG-Output.cpp: In function 'const char* Inkscape::Text::weight_to_text(PangoWeight)': libnrtype/Layout-TNG-Output.cpp:214: warning: enumeration value 'PANGO_WEIGHT_SEMIBOLD' not handled in switch livarot/float-line.cpp: In member function 'void FloatLigne::Affiche()': livarot/float-line.cpp:518: warning: format '%lu' expects type 'long unsigned int', but argument 2 has type 'size_t' livarot/float-line.cpp:524: warning: format '%lu' expects type 'long unsigned int', but argument 2 has type 'size_t' widgets/sp-color-wheel.cpp: In function 'void sp_color_wheel_set_color(SPColorWheel*, const SPColor*)': widgets/sp-color-wheel.cpp:229: warning: operation on 'sat' may be undefined

3 3