Pierre Alexandre Tremblay schreef op 30-08-2023 9:27:
Dear Miller and Christof
I’m sorry to be the cause of pain - I will amend the patch now but what is strange in this one is that recoding it actually changes its behaviour.
In all cases, I wish I had seen the changelog entry that would tell me how the hidden but essential donecanvasdialog settings / behaviour has changed. In all case, I hope this unexpected behaviour report is somewhat helpful.
WHERE IS THIS CHANGELOG ENTRY TO BE FOUND?
ROLF
I know my desire to stay vanilla complicates things too, but again, so many combinations of hardware and oses I think might justify my self-inflicted pain
p
On 29 Aug 2023, at 23:36, Christof Ressi info@christofressi.com wrote:
Sigh.
@Miller Can we please finally get https://github.com/pure-data/pure-data/pull/627? This PR has been sitting around for more than 4 years now. In the meantime, people kept abusing the private "donecanvasdialog" message in place of a better and officially supported alternative, and they will continue to do so...
Christof
On 30.08.2023 00:14, Pierre Alexandre Tremblay wrote:
Ok a lot more investigations later, I have isolated the change of behaviour.
donecanvasdialog 1 -1 0 0 0 0 0 0 0 0 0
Which was used to reset the patcher’s sub patchers at the top left is corrupting the ability for a struct to broadcast its click through a graph. Now, can anyone on the Pd dev confirm this?
How to reproduce:
https://forum.pdpatchrepo.info/topic/10854/xy-abstraction-to-get-mouse-click...
In pd 0.53.2 it works
In pd 0.54.0 it doesn’t receive the click from the #0-rectangle (so it draws and emits nothing)
To fix it, remove top left the donecanvasdialog elements (2 of them)
Then reinstantiate the 3 pd Subpatches on the right
(Or you can just try the one I attach)
It is not fixed completely yet but the click gets through now. I really wonder what has changed there…
Any pointer (pun intended) welcome
p
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
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Oh the capital shout - I hadn’t seen that in a while :)
There isn’t but I won’t blame anyone, they are a pain to write. So much love goes into software maintenance…
I’m spying this:
https://github.com/pure-data/pure-data/commits/master
On 30 Aug 2023, at 10:13, rolfm@dds.nl wrote:
Pierre Alexandre Tremblay schreef op 30-08-2023 9:27:
Dear Miller and Christof I’m sorry to be the cause of pain - I will amend the patch now but what is strange in this one is that recoding it actually changes its behaviour. In all cases, I wish I had seen the changelog entry that would tell me how the hidden but essential donecanvasdialog settings / behaviour has changed. In all case, I hope this unexpected behaviour report is somewhat helpful.
WHERE IS THIS CHANGELOG ENTRY TO BE FOUND?
ROLF
I know my desire to stay vanilla complicates things too, but again, so many combinations of hardware and oses I think might justify my self-inflicted pain p
On 29 Aug 2023, at 23:36, Christof Ressi info@christofressi.com wrote: Sigh. @Miller Can we please finally get https://github.com/pure-data/pure-data/pull/627? This PR has been sitting around for more than 4 years now. In the meantime, people kept abusing the private "donecanvasdialog" message in place of a better and officially supported alternative, and they will continue to do so... Christof On 30.08.2023 00:14, Pierre Alexandre Tremblay wrote:
Ok a lot more investigations later, I have isolated the change of behaviour. donecanvasdialog 1 -1 0 0 0 0 0 0 0 0 0 Which was used to reset the patcher’s sub patchers at the top left is corrupting the ability for a struct to broadcast its click through a graph. Now, can anyone on the Pd dev confirm this? How to reproduce:
https://forum.pdpatchrepo.info/topic/10854/xy-abstraction-to-get-mouse-click...
In pd 0.53.2 it works In pd 0.54.0 it doesn’t receive the click from the #0-rectangle (so it draws and emits nothing) To fix it, remove top left the donecanvasdialog elements (2 of them) Then reinstantiate the 3 pd Subpatches on the right (Or you can just try the one I attach) It is not fixed completely yet but the click gets through now. I really wonder what has changed there… Any pointer (pun intended) welcome p _______________________________________________ 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
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
WHERE IS THIS CHANGELOG ENTRY TO BE FOUND?
HERE ARE THE RELEASE NOTES, WHICH INCLUDE THE MOST IMPORTANT (BUT NOT ALL!) CHANGES: http://msp.ucsd.edu/Pd_documentation/x5.htm#s1
how the hidden but essential donecanvasdialog settings / behaviour has changed.
To be clear: "donecanvasdialog" is an undocumented *private* message. (It is sent from the GUI back to the core upon closing the canvas property dialog.) The behavior may change anytime and it won't appear in a changelog.
If anything, it is better to use the "coords" message instead because it is used in Pd patch files and can be considered stable. However, there are still pitfalls with this approach, so the only sane way is to use a proper public message as proposed in https://github.com/pure-data/pure-data/pull/627.
In all case, I hope this unexpected behaviour report is somewhat helpful.
Yes, it might be still interesting to know what has changed exactly because there might be a regression. Can you come up with a minimal reproducing example and open an issue on GitHub? Thanks!
Christof
On 30.08.2023 11:23, Pierre Alexandre Tremblay wrote:
Oh the capital shout - I hadn’t seen that in a while :)
There isn’t but I won’t blame anyone, they are a pain to write. So much love goes into software maintenance…
I’m spying this:
https://github.com/pure-data/pure-data/commits/master
On 30 Aug 2023, at 10:13, rolfm@dds.nl wrote:
Pierre Alexandre Tremblay schreef op 30-08-2023 9:27:
Dear Miller and Christof I’m sorry to be the cause of pain - I will amend the patch now but what is strange in this one is that recoding it actually changes its behaviour. In all cases, I wish I had seen the changelog entry that would tell me how the hidden but essential donecanvasdialog settings / behaviour has changed. In all case, I hope this unexpected behaviour report is somewhat helpful.
WHERE IS THIS CHANGELOG ENTRY TO BE FOUND?
ROLF
I know my desire to stay vanilla complicates things too, but again, so many combinations of hardware and oses I think might justify my self-inflicted pain p
On 29 Aug 2023, at 23:36, Christof Ressi info@christofressi.com wrote: Sigh. @Miller Can we please finally get https://github.com/pure-data/pure-data/pull/627? This PR has been sitting around for more than 4 years now. In the meantime, people kept abusing the private "donecanvasdialog" message in place of a better and officially supported alternative, and they will continue to do so... Christof On 30.08.2023 00:14, Pierre Alexandre Tremblay wrote:
Ok a lot more investigations later, I have isolated the change of behaviour. donecanvasdialog 1 -1 0 0 0 0 0 0 0 0 0 Which was used to reset the patcher’s sub patchers at the top left is corrupting the ability for a struct to broadcast its click through a graph. Now, can anyone on the Pd dev confirm this? How to reproduce:
https://forum.pdpatchrepo.info/topic/10854/xy-abstraction-to-get-mouse-click...
In pd 0.53.2 it works In pd 0.54.0 it doesn’t receive the click from the #0-rectangle (so it draws and emits nothing) To fix it, remove top left the donecanvasdialog elements (2 of them) Then reinstantiate the 3 pd Subpatches on the right (Or you can just try the one I attach) It is not fixed completely yet but the click gets through now. I really wonder what has changed there… Any pointer (pun intended) welcome p _______________________________________________ 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
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
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
the multitouch xy object https://lists.puredata.info/pipermail/pd-list//2018-10/123713.html this plus the rust app microwave which uses bevy game engine is the only thing ive seen that actually works with multitouch laptop screen on linux , its lije even the apps like surge that have touchscreen mode, the virtual keyboard only treats touches like a single mouse pointer thus only monophonic input. i guess thats a bit off subject, can start a new thread to rant about lack of multitouch input on linux , you would think there were tons of programs for playing polyphonic music and twidlin more than one control at a time in this day and age
On Wed, Aug 30, 2023, 5:40 AM Christof Ressi info@christofressi.com wrote:
WHERE IS THIS CHANGELOG ENTRY TO BE FOUND?
HERE ARE THE RELEASE NOTES, WHICH INCLUDE THE MOST IMPORTANT (BUT NOT ALL!) CHANGES: http://msp.ucsd.edu/Pd_documentation/x5.htm#s1
how the hidden but essential donecanvasdialog settings / behaviour has
changed. To be clear: "donecanvasdialog" is an undocumented *private* message. (It is sent from the GUI back to the core upon closing the canvas property dialog.) The behavior may change anytime and it won't appear in a changelog.
If anything, it is better to use the "coords" message instead because it is used in Pd patch files and can be considered stable. However, there are still pitfalls with this approach, so the only sane way is to use a proper public message as proposed in https://github.com/pure-data/pure-data/pull/627.
In all case, I hope this unexpected behaviour report is somewhat helpful.
Yes, it might be still interesting to know what has changed exactly because there might be a regression. Can you come up with a minimal reproducing example and open an issue on GitHub? Thanks!
Christof
On 30.08.2023 11:23, Pierre Alexandre Tremblay wrote:
Oh the capital shout - I hadn’t seen that in a while :)
There isn’t but I won’t blame anyone, they are a pain to write. So much
love goes into software maintenance…
I’m spying this:
https://github.com/pure-data/pure-data/commits/master
On 30 Aug 2023, at 10:13, rolfm@dds.nl wrote:
Pierre Alexandre Tremblay schreef op 30-08-2023 9:27:
Dear Miller and Christof I’m sorry to be the cause of pain - I will amend the patch now but what is strange in this one is that recoding it actually changes its behaviour. In all cases, I wish I had seen the changelog entry that would tell me how the hidden but essential donecanvasdialog settings / behaviour has changed. In all case, I hope this unexpected behaviour report is somewhat helpful.
WHERE IS THIS CHANGELOG ENTRY TO BE FOUND?
ROLF
I know my desire to stay vanilla complicates things too, but again, so many combinations of hardware and oses I think might justify my self-inflicted pain p
On 29 Aug 2023, at 23:36, Christof Ressi info@christofressi.com
wrote:
Sigh. @Miller Can we please finally get
https://github.com/pure-data/pure-data/pull/627? This PR has been sitting around for more than 4 years now. In the meantime, people kept abusing the private "donecanvasdialog" message in place of a better and officially supported alternative, and they will continue to do so...
Christof On 30.08.2023 00:14, Pierre Alexandre Tremblay wrote:
Ok a lot more investigations later, I have isolated the change of
behaviour.
donecanvasdialog 1 -1 0 0 0 0 0 0 0 0 0 Which was used to reset the patcher’s sub patchers at the top left
is corrupting the ability for a struct to broadcast its click through a graph. Now, can anyone on the Pd dev confirm this?
How to reproduce: >
https://forum.pdpatchrepo.info/topic/10854/xy-abstraction-to-get-mouse-click...
In pd 0.53.2 it works In pd 0.54.0 it doesn’t receive the click from the #0-rectangle (so
it draws and emits nothing)
To fix it, remove top left the donecanvasdialog elements (2 of them) Then reinstantiate the 3 pd Subpatches on the right (Or you can just try the one I attach) It is not fixed completely yet but the click gets through now. I
really wonder what has changed there…
Any pointer (pun intended) welcome p _______________________________________________ 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
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
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
-
Billy Stiltner -
Christof Ressi -
Pierre Alexandre Tremblay -
rolfm＠dds.nl