[LinuxCNC/linuxcnc PR#171] image-to-gcode: compensate for incompatible changes in numpy

未分类 bolang 4个月前 (10-15) 26次浏览

Issue #171 | 状态: 已关闭 | 作者: jepler | 创建时间: 2016-10-05


We use numpy.numarray, which was deprecated in numpy 1.9 and removed in 1.10; Ubuntu 16.04, Mint 18, and Debian 9 have or will have versions without numpy.numarray.


评论 (4)

#1 – jepler 于 2016-10-05

We are already using numpy in all released versions, it’s just numpy.numarray that we need to stop using.


#2 – SebKuzminsky 于 2016-10-05

Looks good. Can you run this through the buildbot to verify that the 2.7 platforms all support numpy.array etc?


#3 – jepler 于 2016-10-05

unfortunately I don’t think we runtest image-to-gcode at all at the moment. in fact there’s no way to run it without a user interface, so doing this kind of testing would require additional changes.


#4 – jepler 于 2016-10-05

Precise and Wheezy have 1.6.1 and 1.6.2 respectively. I could easily test all the new names used on wheezy, and all looks well:

>>> import numpy
>>> numpy.array

>>> numpy.float32

>>> numpy.core.Inf
inf
>>> numpy.fromstring

>>> numpy.zeros

none of these items are listed as new or changed in the release notes for 1.6.2, https://github.com/numpy/numpy/blob/master/doc/release/1.6.2-notes.rst so I think we’re in good shape.


原始Issue: https://github.com/LinuxCNC/linuxcnc/pull/171

喜欢 (0)