Thanks for the feedback - adding the connect parameter does appear to make
it behave properly.
Playing with it again on my full application (without the additional
parameter), it looks like it happens more frequently when the (Y) axis is
manually scaled.
Should I open this as a Github issue?
On Saturday, 1 October 2022 at 06:36:07 UTC+1 Patrick wrote:
> Can reproduce with latest version of pyqtgraph. Some bug with detection of
> nan/infs, but only on subsequent updates (note the first draw renders
> correctly). Adding an explicit connect="finite" to the setData calls works
> as expected.
>
> On Saturday, 1 October 2022 at 12:57:30 am UTC+9:30 AB wrote:
>
>> OK, I've hacked down the application to a small-ish file (attached).
>> There are two graphs, side-by-side. They both have some data with an
>> invalid section and it is consistently being drawn across until I change
>> the zoom level - it then reappears on the next graph update.
>>
>> On Friday, 30 September 2022 at 16:11:44 UTC+1 AB wrote:
>>
>>> np.nan also misbehaves.
>>>
>>> On Friday, 30 September 2022 at 16:09:38 UTC+1 AB wrote:
>>>
>>>> Since posting earlier, I've tried float("nan"), float("inf") and np.inf
>>>> and they all behave exactly the same. I'll see if I can cut my
>>>> application's code down to something I can share.
>>>>
>>>> I did manage to "stop" the graph while the glitch was displayed (and by
>>>> "stop" I mean that the application was running but I stopped making
>>>> further
>>>> calls to setData). Before zooming the graph I used the context-menu to
>>>> export data to CSV and I could clearly see that there was a period
>>>> exported
>>>> as "inf" where the line was being displayed.
>>>>
>>>> On Friday, 30 September 2022 at 16:02:50 UTC+1 [email protected]
>>>> wrote:
>>>>
>>>>> Hmm, that certainly should not be happening, not sure quite why things
>>>>> would be going like that. If you can, I would try slowing down the
>>>>> update
>>>>> rate of the plot, to easily catch one of those instances and post some
>>>>> code
>>>>> here if you can for us to try and replicate.
>>>>>
>>>>> Another suggestion would be to set the invalid points to numpy.nan
>>>>> instead of float("inf") and see if that gets better behavior; I think
>>>>> that
>>>>> was the value we were originally looking for.
>>>>>
>>>>> On Fri, Sep 30, 2022 at 7:27 AM AB <[email protected]> wrote:
>>>>>
>>>>>> I am trying to display some data in a line graph - there are 1024
>>>>>> points, updating ~40 times/second. Some of the points are invalid (set
>>>>>> to
>>>>>> float("inf") ) and are supposed to cause a break in the line if I
>>>>>> understand the default behaviour of the "connect" parameter correctly.
>>>>>> However, I am periodically seeing that there is a straight line joining
>>>>>> the
>>>>>> valid points either side of the invalid period.
>>>>>>
>>>>>> I can stop updating the graph and catch this happening - as soon as I
>>>>>> use the mouse zoom functionality to change the scale of the graph in
>>>>>> this
>>>>>> state the erroneous line vanishes, and the graph is drawn with breaks in
>>>>>> the line as I expect that it should.
>>>>>>
>>>>>> Can anyone suggest why the erroneous lines are appearing, and how I
>>>>>> can ensure they don't appear?
>>>>>>
>>>>>> --
>>>>>> You received this message because you are subscribed to the Google
>>>>>> Groups "pyqtgraph" group.
>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>> send an email to [email protected].
>>>>>> To view this discussion on the web visit
>>>>>> https://groups.google.com/d/msgid/pyqtgraph/559c2c62-8372-471f-b36c-77e314368006n%40googlegroups.com
>>>>>>
>>>>>> <https://groups.google.com/d/msgid/pyqtgraph/559c2c62-8372-471f-b36c-77e314368006n%40googlegroups.com?utm_medium=email&utm_source=footer>
>>>>>> .
>>>>>>
>>>>>
--
You received this message because you are subscribed to the Google Groups
"pyqtgraph" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/pyqtgraph/0013f7d6-5daa-4157-80e8-0c1f5b861669n%40googlegroups.com.
