Coverage decreased (-0.2%) to 75.764% when pulling 8e6a853 on ochurlaud:patch-1 into 7b07af3 on python-control:master.

I would not expect this to only affect Python 3.

Can you provide an example that demonstrates the error?

Since np as an identifier for numpy is not used within the function scope after where np is assigned to an array size, it is not an error. I agree that changing the name would improve the style, but it is not a matter of correctness.

Hum, I don't remember: it was some time ago. Comparing to other modules, t should be call _np I think instead of what I wrote. Anyway: as it's a small fix, without huge repercussions, I think it should be somehow done.

Even if you don't use my PR, just do the correction somehow?

I fixed this on #100. It is a bug; if, on that branch, I revert from np_ to np, I get the error below. I'd have to read up on how Python handles globals and locals to explain this (did that change in Python 3?).

Test h2syn
----------------------------------------------------------------------
Traceback (most recent call last):
  File "robust_test.py", line 34, in testH2syn
    k = control.robust.h2syn(p, 1, 1)
  File "/home/rory/src/python-control/control/robust.py", line 91, in h2syn
    n = np.size(P.A,0)
UnboundLocalError: local variable 'np' referenced before assignment

Ok, then maybe merging #100 would be better than mine.

Changes from pull request #100 were merged, which subsumes changes from here, as noted above. In particular, cf. lines 93 and 157 of control/robust.py as of commit d4e0155 (current tip of master branch).

@ochurlaud Nonetheless, thanks for this pull request. I should have been faster to review it.