The change in drawValueStaircase from

if (poly_x.size() > 0  && x != last_x)

to

if (poly_x.size() > 0  && x != last_x && !Double.isNaN(value))

causes a problem.

Consider a trace with disconnected data.

The archived samples are considered to be valid until there's a new sample. This includes the disconnected case. The trace, at least in "staircase" mode, should continue from the last known sample to the point of disconnect.
After this change, the trace stops at each sample before the disconnect, resulting in large gaps:

By restoring drawValueStaircase to use just if (poly_x.size() > 0 && x != last_x) as was the case before this change, the trace continues up to the disconnect, resulting in only a brief gap that more correctly represents the value over time:

One would assume that Stripchart and Databrowser uses the same plot engine...

Yes. Your original screenshot suggests that the order of samples is somehow getting disturbed. The time stamps must be sorted. The 'fix' on the plotting side just added a new bug.

OK, will take a look.

Since I cannot reproduce the issue (and neither can the user reporting on it), I have reverted this change.
See #3736