Hi,
I just posted a bug report about this, but I am in an urgency of finding a quick workaround.
On Windows XP, sometimes, when scrolling up and down a big patch window with a huge lot of gui stuff (many instances of a few gop-enabled abstractions), USING THE MOUSE SCROLLWHEEL, I get a "Divide by Zero" error dialog, which does NOT crash or hang PD (but does annoy):
""" divide by zero divide by zero while executing "expr -abs(0)/0" invoked from within "pdtk_canvas_scroll .x9a4e60.c y [expr -abs(0)/0]" (command bound to event) """
It never happened to me before, so I suspect it happens when the patch you're scrolling is very "big" (in terms of quantity of gui elements). Seems to happen only with the scroolwheel, not the scroll bar.
If I scroll a lot up and down and up and down, I am almost sure to reproduce it within less than a minute.
I have been working for months developing this patch, which is going to be used on stage in a show the day after tomorrow and suddenly I add a couple of instances and this bug appears; though it does not crash pd it does worry me, so my question is: does anybody know this bug and know whether there are some particular circumstances triggering it so that maybe I can avoid them for the moment??
(avoiding scrollwheel scrolling is just not feasible, as you can even move the mousewheel unintentionally, furthermore the sound technician managing the audio through the patch needs to be able to scroll in a handy way and has always done it with the scrollwheel since it was never a problem up to now)
PD Vanilla 0.40.2 on Windows XP. Clean install with no externals.
Any help will be appreciated Thanks m.
Please post a bug in the tracker. Attach a patch that this happens
with, and the instructions to reproduce the bug. The easier it is to
reproduce this bug in your patch, the easier it will be to fix the bug:
http://puredata.org/dev/bugtracker
.hc
On Oct 22, 2007, at 12:06 PM, matteo sisti sette wrote:
Hi,
I just posted a bug report about this, but I am in an urgency of finding a quick workaround.
On Windows XP, sometimes, when scrolling up and down a big patch window with a huge lot of gui stuff (many instances of a few gop-enabled abstractions), USING THE MOUSE SCROLLWHEEL, I get a "Divide by Zero" error dialog, which does NOT crash or hang PD (but does annoy):
""" divide by zero divide by zero while executing "expr -abs(0)/0" invoked from within "pdtk_canvas_scroll .x9a4e60.c y [expr -abs(0)/0]" (command bound to event) """
It never happened to me before, so I suspect it happens when the patch you're scrolling is very "big" (in terms of quantity of gui elements). Seems to happen only with the scroolwheel, not the scroll bar.
If I scroll a lot up and down and up and down, I am almost sure to reproduce it within less than a minute.
I have been working for months developing this patch, which is going to be used on stage in a show the day after tomorrow and suddenly I add a couple of instances and this bug appears; though it does not crash pd it does worry me, so my question is: does anybody know this bug and know whether there are some particular circumstances triggering it so that maybe I can avoid them for the moment??
(avoiding scrollwheel scrolling is just not feasible, as you can even move the mousewheel unintentionally, furthermore the sound technician managing the audio through the patch needs to be able to scroll in a handy way and has always done it with the scrollwheel since it was never a problem up to now)
PD Vanilla 0.40.2 on Windows XP. Clean install with no externals.
Any help will be appreciated Thanks m.
-- Matteo Sisti Sette matteosistisette@gmail.com http://www.matteosistisette.com
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
I spent 33 years and four months in active military service and
during that period I spent most of my time as a high class muscle man
for Big Business, for Wall Street and the bankers. - General
Smedley Butler
hi
I just posted a bug report about this, but I am in an urgency of finding a quick workaround.
Hans-Christoph Steiner wrote:
Please post a bug in the tracker.
actually matteo just wanted to point out that he has already submitted a bug-report to the bug-tracker [1817858]
i think he only wanted to raise the priority on a social level. however, if it is so important for him, he should have noticed that 45minutes after he had submitted the bug, i have already posted a possible solution.
mfga.dr IOhannes
actually matteo just wanted to point out that he has already submitted a bug-report to the bug-tracker [1817858]
i think he only wanted to raise the priority on a social level.
I am really sorry if I annoyed anyone with the double post, but posting the issue to the bug tracker and to the mailing list had two different purposes and was addressed to different (though obviously overlapping) audiences. The bug tracker is intended for developers to let them know the bug exists (in case they don't already) so that they can fix it some day; and since I searched the bug tracker and didn't find it reported (which doesn't necessarily mean it has never been reported), I felt the duty to do so. Otoh, a message to the list was addressed to users, who may help me to work the issue around without waiting for a next version with a fix.
If I didn't wait even 45 minutes (nor even 5 probably) after reporting the bug before I wrote to the ml, it was not for impatience but just because the two messages were intended for different targets: just like one may post a question to two (or more) different forum all related to the topic of his question. I didn't even expect any quick workaround or "hack" to be posted to the tracker.
Thank you very much IOhannes for the fix you posted, however I'm afraid I can't try it out. Correct me if I'm wrong: I need to recompile PD after modifying the source code with the patch, right? I don't have the skills to do that right now, sorry.
Btw, I am using Windows version of PD (vanilla 0.40.2), sorry I didn't mention it in the report (i usually do, it was an oversight).
Thanks, m.
Matteo Sisti Sette wrote:
actually matteo just wanted to point out that he has already submitted a bug-report to the bug-tracker [1817858]
i think he only wanted to raise the priority on a social level.
If I didn't wait even 45 minutes (nor even 5 probably) after reporting the bug before I wrote to the ml, it was not for impatience but just because the
oh sorry, i was not aware of that. to me it seemed that between your posting to the tracker and to the ml a timespan of 4 hours elapsed. (the tracker-item was created at 14:42 to which is responded at 15:23, my mailclient (thunderbird) shows the _sending_ date for your email as 18:06CEST) i admit i haven't fully checked whether your email was stuck in some greylisting/... and therefore whether it was actually sent at about the same time....
two messages were intended for different targets: just like one may post a question to two (or more) different forum all related to the topic of his question. I didn't even expect any quick workaround or "hack" to be posted to the tracker.
darn, i am too fast...
Thank you very much IOhannes for the fix you posted, however I'm afraid I can't try it out. Correct me if I'm wrong: I need to recompile PD after modifying the source code with the patch, right?
wrong. you just need to edit the pd.tk file which is read and interpreted at runtime by the tcl/tk-interpreter (wish on w32) the only thing you have to do is restart pd...
I don't have the skills to do that right now, sorry.
it is very simple: download the patch and open it with a decent text-editor (notepad might have problems with the carriage returns) any line prefixed with a "-" is to be removed from the original source, and line prefixed with a "+" is to be added. the lines without prefixes are there to give you a context (e.g. if you have already applied another patch (e.g. because you are not using pd-vanilla's pd.tk) the line-numbers might differ, but the context will hopefully be the same)
what you basically have to do is change the 2 occurences of "-%D/abs(%D)" to "(%D>0)?:-1:1" (or something like this, i don't remember exactly)
Btw, I am using Windows version of PD (vanilla 0.40.2), sorry I didn't mention it in the report (i usually do, it was an oversight).
no big problem; i imagined something like this :-) the problem can only occur on w32 and os-x (linux/freebsd uses different code here that cannot divide by zero)
mfga.dr IOhannes
IOhannes m zmoelnig wrote:
Matteo Sisti Sette wrote oh sorry, i was not aware of that. to me it seemed that between your posting to the tracker and to the ml a timespan of 4 hours elapsed. (the tracker-item was created at 14:42 to which is responded at 15:23, my mailclient (thunderbird) shows the _sending_ date for your email as 18:06CEST)
which i have to admit does not sum-up to 4 hours (if the sf-time is really given in local time, which i don't know)
fgmas.dr IOhannes
On 22/10/2007, at 18.06, matteo sisti sette wrote:
It never happened to me before
It did happen to me before, cf. [1]. But it seams to be "fixed" in
later versions of Pd-extended-0.39.x
As i describe in that email it happened all the time when i tried to
scroll or had scrolling enabled.
[1] http://lists.puredata.info/pipermail/pd-list/2006-06/039351.html
It did happen to me before, cf. [1]. But it seams to be "fixed" in later versions of Pd-extended-0.39.x
I'm reproducing it in PD-Vanilla 0.40.2 on Windows XP.
I don't know what the "scroll with two fingers" option is, however i get the error (only sometimes, at random times) when scrolling when the mouse wheel, without even touching the touchpad.
-
Hans-Christoph Steiner -
IOhannes m zmoelnig -
IOhannes m zmölnig
-
Matteo Sisti Sette -
matteo sisti sette
-
Steffen Juul