Wrong graphical representation with mode=6 and lsize>1 and many datapoints


I am evaluating data and test different versions of idealization. For this I plot data and idealization as cityscape (mode=6). To judge the idealization visually, I make it thicker ( lsize = 2 or 3).

Now depending on the zoom the thick lines have amplitudes quite a bit off ( see time: 13.18).

Example is attached, in the zoomed out screen grab the grey (idelaized line) extends below the red line ( which the data does not). Zoomed in , the data is shown correctely. Also, when saved as png the data is correct.

My guess is, that this is an effect showing when there are more data-points than screen pixels.

Are there any display setting to fix this?

I wasted quite some time as I thought my routine calculated bs...


Any help would be great.





Ps: This is Igor on Win7



Grey line extend below red. Data does not.

I have problems getting all 3 graphs to display, therefor a zipfile with the pngs and a pxp of the graph.

Interestingly I did not had the problem when I started this project still under Igor 6.*. I just confirmed it on another PC:

The graph is displayed correctly in Igor 6, but not in Igor7.

Unfortunately going back to igor 6 is not an option, as I do not have a 32 bit matlab available, as I load my data from mat files....

Again: Any workaround would be highly helpfull.






The bug is fixed for the next nightly build of Igor 7 and 8. Tomorrow you should be able to get a fixed Igor 8 from the nightly build. Use the Help->Igor Pro Nightly Build menu item or visit https://www.wavemetrics.net/Downloads/latest8/ directly to get the latest version. Make sure you get a build number at least 32961.

The odd bug was caused by a failure to set the line join style, and the default was a miter join with miter limit of 10. The odd extensions of zero-width cityscape blocks was the line join miter, which for two lines meeting at 360 degrees should be infinite. Be glad we limited it to 10! For Igor 7, you have to accept our judgement for the best default, which is mitered with a miter limit of sqrt(2), which makes 90-degree joins have a nice square corner.

In Igor 8, the join style, miter limit and cap style are all under your control. That means you could re-introduce the bug if you want :)