Mon 17 Dec 2012 12:16:48 AM UTC, comment #9:
Ok. I pushed the change.
http://hg.savannah.gnu.org/hgweb/octave/rev/d0579fed7e22
|
Mon 17 Dec 2012 12:08:43 AM UTC, comment #8:
No objection. Thanks
|
Sun 16 Dec 2012 11:39:40 PM UTC, comment #7:
I'm attached a changeset and a modified version of Pantxo's hdlcmp.m function. I've added a demo script to hdlcmp.m.
With the attached changeset, the resulting copies created using copyobj() essentially identical. The only difference is with the tick label positions where are off by less than eps().
Any objection to push?
(file #27109, file #27110)
|
Sat 15 Dec 2012 04:45:47 PM UTC, comment #6:
I can confirm that listeners are triggered only if the value is changed regardless of the graphics_toolkit:
|
Sat 15 Dec 2012 04:11:05 PM UTC, comment #5:
Pantxo, my understanding is that the listeners are only triggered if the value of the property is changed. Thus, the command below should not trigger a listener for the "position" property.
If it does, then something unintended is happening.
For the gnuplot graphics toolkit, the tick values of 0.6 are off by 0.5 eps (). Looks like there is a small numeric problem somewhere.
I added the demo below to your hdlcmp.m to see the size of the difference in the property values.
Regarding the fltk graphics toolkit, the problem may be independent of copyobj(). Perhaps Konstas will have some insight into this. I've cc'd him.
|
Sat 15 Dec 2012 01:57:00 PM UTC, comment #4:
I have added a listener to position and ztick properties to see how much times they are modified after the axes has been created.
If I run the example below I can see that in gnuplot ztick is modified once, as expected, but position is never modified. In fltk ztick is also modifed once but position is modifed 15 times.
Wild guess: in gnuplot position is updated only if the requested value is different from the current one?
|
Sat 15 Dec 2012 01:13:38 PM UTC, comment #3:
My tip is:
I use the attached hdlcmp.m to compare model and clone properties. The example below still shows me what I explained previously.
(file #27100)
|
Fri 14 Dec 2012 11:31:39 PM UTC, comment #2:
My tip is;
The example works for me. Was this bug fixed by the recent changesets to hdl2struct()? For example, the one below.
|
Thu 13 Dec 2012 11:05:42 PM UTC, comment #1:
I can confirm on linux mint. I have different behaviors for gnuplot and fltk :
- gnuplot : "x(yz)tick" properties are slightly different. The axes look perfectly matched though.
- fltk : same as gnuplot plus "tightinset" (read only) and "position" that are clearly different (and so does the axes look). If I force he position afterward, the axes seem to match but the tick labels don't.
|
Thu 13 Dec 2012 11:05:17 AM UTC, original submission:
I one does
the two plots do not match.
|