Not a bug report

I just tested 1_wrap on my setup and it runs fine, however I get a following error with 1.ngc (on both 2.8 and 2.7)

Since this behavior is very setup dependent, can you _please_ plot the following
error on your configuration so developers have something to work with...

EDIT
Here are a couple example plots:

freeby.mesanet.com/28ferror.png
freeby.mesanet.com/28ferrorb.png

The acceleration limit on my machine is 200 (the acceleration reaches almost 1000)

I have stayed with 2.7.14 and it works well, in fact, flawless as previously stated.
I am not daft enough to fix something that isn't broke but it seems that's exactly what has happened with the upgrade.
Whether recommended or not, if the new versions are not tested thoroughly we would have a substandard system which the current version is not but the upgrade will be if released too soon.
I have no intentions of fault finding or effecting fixes, that's for the developers to do but I can assist by pointing out any issues I may have and supply the goods for them to replicate my issue.
The issue was not present in all previous versions of that mentioned above but remains in the current version.

The configs you supplied are not working parallel port configurations they are both sim configs.

Note that 1_wrap.ngc works with LinuxCNC 2.8 for me both on real hardware and on
sim configurations. This is why providing actual data (like an error plot) can help find the issue, as it may be nearly impossible to duplicate your exact setup that triggers the error without some hint of what the actual error is...

I have worked with this a bit more and at least in my case i can fix all acceleration violations
I have found by setting the G64 PXXX value to a reasonable number.

Both of your example GCode files are in mm and have G64 P 0.001 (1 micron blending!)
If I change this to 0.01 mm blending I no longer have acceleration violations in 1.ngc
with my (very aggressive) velocity and acceleration settings

But my point is: I don't need to fix anything in the current release nor in the earlier versions of 2.8.
And yes they are sim configs as I had to put the parport machine back to version 2.7.14 after the upgrade didn't work.
The 0.001 comes from the drawing line thickness as I am engraving jewelry and need to work within extremely fine tolerances.

Thank you for your input. I will stick with the stable version until the final release of 2.8.
The beauty of Linuxcnc is it can be installed completely in a few minutes unlike the alternatives, so it's easy to backpedal a version or two if needed.

2.7 has the exact same bug as 2.8 with too small PXXX values
I would suggest trying reasonable values for PXXX to see if it fixes the issue
(.001 mm is not a reasonable number here considering that your actual step resolution is only .0025 mm)

Whatever it was it has been sorted now and all is working well again, till the next time it breaks.
Just as well because I don't fancy changing thousands of ngc files to suit a glitch in the system.