On Sat, 2016-04-30 at 12:08 -0700, Miller Puckette wrote:
OK.. I added a 0.46 compatibility mode to get the old, strange behavior :)
I don't see anything regarding this in the logs. Did you push? The most recent commit I see is adcdcede.
Now I feel somewhat guilty for another compatibility mode ;-) Probably more important than supporting the 'strange' behavior would be to reflect the current behavior in the documentation. See attached my_canvas-help.pd
Roman
On Sat, Apr 30, 2016 at 05:42:55PM +0200, Roman Haefeli wrote:
On Fri, 2016-04-29 at 15:49 -0700, Miller Puckette wrote:
It looks like the new color handling has made teh "color" message work more coherently. In the past, if "color" had two arguments they were interpreted as background, foreground, and label color, but if only two were supplied theyt were the foreground and label colors (without setting background color).
Actually, a color message with two arguments sets background and label, leaving foreground color unchanged (in Pd <= 0.46).
I think the old behavior is horribly confusing.
Until now, I wasn't aware that all iemguis behave the same regarding colors. The new behavior would have worked for old versions of Pd, but it never occurred to me, since the documentation for [cnv] clearly suggests using only two arguments (background, label), while all other iemguis require three arguments. Thus, I used only two arguments whenever setting the label color for [cnv]s dynamically. The change of behavior breaks quite a few patches of mine. This doesn't mean I'm opposed to introducing sanity in that matter (it's quite easy to fix), but I'm wondering how many other patches with non-trivial GUI designs would be affected.
I'm not sure whether to consider it a bug or a 'feature' - and whether back compatibility is more important than making the color-settings sane.
I'm not even sure what is more sane. The new behavior forces the use of a redundant middle argument when setting the label of a [cnv].
Roman
On Mon, Apr 25, 2016 at 10:47:38PM +0200, Roman Haefeli wrote:
I'm glad to see the release that includes deken is out now.
One minor thing still apparent in the test2 release: The second argument of the 'color' message to a [cnv] object does not change the text color anymore in Pd 0.47. Text color can be changed by providing a _third_ argument, though.
See attached patch.
Cheers and thanks for all the work gone into that release, Roman
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list