> > > I also added a warning when no entities are replaced within the icon. > Do you feel that I should actually increase the granularity, instead > providing warnings separately for strokes and fills? >
I chose to separate these, so that it will warn that no strokes were replaced even if some fills were, and vice versa. > The only remaining "oddity" in the script right now is that, > regardless of the colors that are used for stroke/fill entities, the > script will set the stroke and fill colors in the output to (#666666, > #ffffff) when a guess is made. On the other hand, when you explicitly > pass them with -s and -f, the converted icon will retain the > explicitly set colors. What do people think correct behavior is here? > Always resort to the default color pair, even when the colors are > passed? (You can just edit the entity declarations at the top to set > it to anything after running the script...) Always use the passed or > guessed values? (I didn't do it this way because, due to some > restrictions in the way the SVG DOM can be edited, I have to insert > the entities into the raw text before parsing the DOM, which means I > have to do it before I can make a guess as to the correct entity > colors... > I've taken care of this, as it bugged me and seemed inconsistent. The script will now output a sugarized icon that looks exactly the same as the input, by default. If desired, you can pass the -d flag to insert the default color entities that will be recommended for "uninstantiated activity icons". Note that the output colors are therefore independent of the colors passed in with -s and -f, which are used solely to tell the script which hex values to replace with the entities, and not what color to output. I considered adding -S and -F options which would specify the output colors, but I didn't really think of a use case for this...it seems that the input colors or the defaults should be sufficient. Finally, I noticed after my experiments with the Moon icon that the guess algorithm was completely useless, since it used a mask (it always guessed black and white). To get around this, I now filter out mask elements when performing the guess, so it works just fine and makes a valid guess with any number of masks. After the above changes and another batch of bug fixes and robustness improvements, mostly with respect to the -m capability, I'm finally satisfied enough with this to write it up on the wiki. I intend to do that later this afternoon; feel free to try the attached script at your whim. Thanks for your feedback again! - Eben
sugar-iconify
Description: Binary data
_______________________________________________ Sugar mailing list Sugar@lists.laptop.org http://lists.laptop.org/listinfo/sugar