yeah, changing the 2 to 3 always works but the remaining problem is: what about all those patches which have a '2' in the [donecanvasdialog( message? I know, dynamic patching is not 'officially' supported but it's a very common practice with whole libraries relying on it.
so is it really a good idea to break those patches (and leave people wondering about the reason) for fixing a bug which IMHO is only a minor inconvenience? are there other possibilities to fix this bug without introducing a regression? if this bug fix makes it to the release than there should be at least a big fat warning in the change logs and other places to give people the chance to update their patches. to make things more confusing: [coords ...( still accepts the flag value 2 for drawing the GOP (without text)... see attached patch. my humble suggestion is to undo the bug fix and think about it/discuss it for a next release. Christof > Gesendet: Montag, 24. Juli 2017 um 15:43 Uhr > Von: ro...@dds.nl > An: "Pd list" <pd-list@lists.iem.at> > Betreff: Re: [PD] Pd 0.48.0test1: donecanvasdialog changed? > > again thanks Christof, > > your diagnosis was correct. > changing the GOP parameter in the donecanvasdialog to 3 (was 2) did the > trick. > > it's also backwards compatible with Pd-extended, which in my case is > necessary. > > rolf > > ro...@dds.nl schreef op 23-07-2017 22:05: > > thanks Christof, > > for the answer > > > > attached the correct patch > > > > rolf > > > > Christof Ressi schreef op 23-07-2017 21:26: > >> Hi, I think something is missing in your patch. what is [waveform-r]? > >> > >> Anyway, I spotted a difference: > >> > >> /* hack - graphme is 0 for no, 1 for yes, and 3 for > >> yes-and-hide-text*/ > >> if (!(graphme & 1)) > >> graphme = 0; > >> > >> In the array dialog you have a checkboxes: "graph-on-parent" and "hide > >> object name and arguments". > >> > >> What happens I think is this: > >> > >> the first checkbox sets the first bit and the second checkbox the > >> second bit in a flag. > >> so 2 only means hide-text but until now it was assumed that you also > >> want to graph it. the correct flag value for graph-and-hide-text would > >> be 3. > >> > >> I think Miller wanted to fix the following bug in the canvas dialog: > >> > >> when you have both checkboxes ticked and you untick the first one > >> (graph-on-parent) the second one stays ticked so the graph still gets > >> drawed. you must untick the second one before you can untick the first > >> one. > >> > >> @Miller: this bug fix will break many patches which rely on > >> [donecanvasdialog(, please revert to the original behaviour! I think > >> the issue you were trying to fix should be handled on the Tcl/Tk side > >> before canvas_donecanvasdialog gets called. > >> > >> Christof > >> > >>> Gesendet: Sonntag, 23. Juli 2017 um 20:39 Uhr > >>> Von: ro...@dds.nl > >>> An: "Pd list" <pd-list@lists.iem.at> > >>> Betreff: Re: [PD] Pd 0.48.0test1: donecanvasdialog changed? (was: > >>> strange behavior graph in struct ) > >>> > >>> i'm using the donecanvasdialog to create a costumized view of a > >>> graph. > >>> > >>> up to Pd 0.47 this works, in Pd 0.48.0test1 not anymore. > >>> > >>> attached a patch which demonstrates this. > >>> > >>> first try the patch with older Pd version to see what it should give, > >>> > >>> then do the same in Pd 0.48. > >>> > >>> > >>> the only difference is the (effect of) donecanvasdialog. > >>> > >>> did it change? > >>> > >>> rolf > >>> > >>> _______________________________________________ > >>> Pd-list@lists.iem.at mailing list > >>> UNSUBSCRIBE and account-management -> > >>> https://lists.puredata.info/listinfo/pd-list > >>> > > _______________________________________________ > Pd-list@lists.iem.at mailing list > UNSUBSCRIBE and account-management -> > https://lists.puredata.info/listinfo/pd-list >
donecanvasdialog-vs-coords.pd
Description: Binary data
_______________________________________________ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list