GNOME Bugzilla – Bug 59881
resize of grouped objects is broken
Last modified: 2009-01-29 21:38:49 UTC
actually, it isn't implemented. Group resizing is actually a difficult problem for structured graphics. Quite often you can't just scale everything affinely.
I don't understand why there is more complexity in resizing 2 objects than there is in resizing 1, apart from an added iterator? All the difficult issues, such as deciding how to resize fonts and line-widths, also apply to single object resizing, so presumably those are already solved?
How do you resize the following sample ? OK, easy. Now, in place of circles, think about UML Classes. How do you resize that diagram ? (and I'm sure James has more ideas about problemful situations). Solving these problems (defining the correct behaviour) would help a lot towards implementing the solution.
I don't understand. Are you telling me that UML shapes are not defined by SVG in .shape files? .....Aghh! I see they are not. Any object, or set of objects, or grouped set of objects, or set of grouped objects, ... , needs to be resizable. So long as every component has an SVG description why would this be difficult? I'm worried that dia is backing itself into a corner here. My recommendation is that dia uniformaly require all shapes to be based on .shape files.
Mass reassign of bugs to dia-maint@bugzilla.gnome.org.
Thanks for the recommendation. Of course, we would then loose all the specific behaviour which makes dia a diagram editor, not a vector program (well, there has been proposals to throw massive doses of XML magic at this problem, but unfortunately for the moment they're still mostly vapourware). The problem is not "make this XML and the problem vanishes". The problem is specifying what the expected _behaviour_ and then implementing it. XML has never been a substitute for engineering (well, ahem, it looks like it has been used as a substitute, in the real world. Whoops.).
This seems as if it is a duplicate of bug 59880 bug 59880 Summary: "Need resize of simple objects, of group objects, of arbitrary selections of objects and/or groups. Need fixed aspect resize,but variable aspect resize could be useful too." if this bug does get closed as duplicate please the copy the most useful comments to bug 59880 if it is not a duplicate please explain.
damn, i dont have enough permissions to change the summary, i was trying to change it as follows: old summary: resize of grouped objects is broken new summary: [RFE] resize of grouped objects wanted I would really like to have groups that are resizalbe for at least the simple cases, circles, polygons, etc. For shapes that contain text or something else that makes them difficult to resize you would pop an error message of some kind. Perhaps Dia should implement the concept of "Locked shapes" that I have seen in Visio and GoBe Productive. I think this feeds back into bug # 95516 because if grouping were available users would be able to ungroup an object remove the unresizable part (eg a bit of text) resize it and add their own text label.
I gave some serious thought to how this is supposed to work. (and then i looked at visio, which i should have done in the first place and saved myself a lot of thinking). Resizing groups Groups of simple shapes resize just the same as a single simple shape would resize. Resising a group that contains text merely resizes the text area / text box and leaves the text alone. The Text widget in Dia does not exaclty lend itself to doing this. I would suggest either leaving it alone and resizing the other objects or instead pop up an error message and refuse to resize groups that contain text objects (or other weird objects like UML). [While i am at it i may as well note a few other things: Groups in visio (unlike in Dia) do not use a different selection colour, they use the same colour green handles.
sorry, this wont happen anytime soon, setting to enhancement to get it out of my way while fixing 'real' bugs ;)
after years I fail to see the difference to the broader bug #59880 *** This bug has been marked as a duplicate of 59880 ***