Release 0.3.5

@murrayrm

I'd like to release current master, or something close to it, as 0.3.5. Our current PyPI release, 0.3.4, is arguably broken: pip install slycot doesn't work on Linux systems (I suspect also not on OSX), but 0.3.5 should fix that.

@murrayrm, could you please test the build on OSX? Something like pip install git+https://github.com/python-control/Slycot in a clean virtual-env (no numpy, no scikit-build, up-to-date pip) should be sufficient.

AFAIK, conda builds are still OK. @repagh, @moorepants, please let me know if there is anything conda-related that must be done.

Current master builds fine on the OpenSUSE build system now, thanks for the fixes!
https://build.opensuse.org/package/show/home:bnavigator/python-slycot

One notes is that we should change build.sh and bld.bat in the conda-forge feedstock to the recommended:

python -m pip install --no-deps --ignore-installed .

Then pip install-ability will be checked downstream. You can do something similar on the Travis builds here.

@moorepants

@moorepants thanks, that's a good idea. From googling (e.g. conda-build docs and a conda-forge tutorial) it looks like one should add pip to the list of host requirements in that case.

Is there any somewhat-authoritative source for this recommendation? Would like to put it in the commit log.

current conda-forge docs recommend "{{ PYTHON }} -m pip install . --no-deps -vv"

The "--ignore" makes sense, so I'll keep that; I'll add -vv.

I'd use the conda-forge docs as the primary source for the conda-forge feedstock and the conda-build for anything you are doing in this repo. Conda-forge has some requirements that are specific to their build infrastructure.

This is what I was looking for originally:

{ PYTHON }} -m pip install . --no-deps -vv

I found python -m pip install --no-deps --ignore-installed . instead. The docs are too confusing! Go with the one you found.

I'd use the conda-forge docs as the primary source for the conda-forge feedstock and the conda-build for anything you are doing in this repo.

My understanding of why we're doing conda-builds at all in our CI is as early-warning of problems for conda-forge, so we might as well follow conda-forge practices.

Speaking of which, is there any way to use the conda-forge recipe in the Slycot, rather than using our own? It feels like we're duplicating effort i.t.o. of conda-build recipe maintenance otherwise.

The conda recipes in the Slycot repo predated me building it on conda forge. I'm not sure I've ever looked at them. If you want to build the conda forge recipes before a release, then you may have to set up a similar CI configuration. It is possible to have conda forge build prereleases from a sub-channel which is another way to trigger a conda forge build on their system and avoiding people installing a new release.

Report on OSX build:

Trying to build from a completely clean environment (Python 3.7, using conda) fails because of the lack of a FORTRAN compiler.
After installing a FORTRAN compiler (using conda install gcc), installation was successful. Output attached below.

(slycot-test) nemaya:~ murray$ pip install git+https://github.com/python-control/Slycot
Collecting git+https://github.com/python-control/Slycot
  Cloning https://github.com/python-control/Slycot to /private/var/folders/mf/mg10yhgs0_j0wr03c7vw0hnc0000gn/T/pip-req-build-n9pvsf1m
  Running command git clone -q https://github.com/python-control/Slycot /private/var/folders/mf/mg10yhgs0_j0wr03c7vw0hnc0000gn/T/pip-req-build-n9pvsf1m
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
    Preparing wheel metadata ... done
Collecting numpy (from slycot==0.3.4.23)
  Using cached https://files.pythonhosted.org/packages/6b/be/608b7f72b851472388eafc010a5d46dae5d41610d0ac5df4c98c2ed1b865/numpy-1.16.4-cp37-cp37m-macosx_10_6_intel.macosx_10_9_intel.macosx_10_9_x86_64.macosx_10_10_intel.macosx_10_10_x86_64.whl
Building wheels for collected packages: slycot
  Building wheel for slycot (PEP 517) ... done
  Stored in directory: /private/var/folders/mf/mg10yhgs0_j0wr03c7vw0hnc0000gn/T/pip-ephem-wheel-cache-1rfdji3s/wheels/56/5f/0e/f52a52e6f7bb57a8d00c45721e2ed8b8f461c936f0f2d2d000
Successfully built slycot
Installing collected packages: numpy, slycot
Successfully installed numpy-1.16.4 slycot-0.3.4.23
(slycot-test) nemaya:~ murray$ pwd
/Users/murray
(slycot-test) nemaya:~ murray$ python
Python 3.7.3 (default, Mar 27 2019, 16:54:48) 
[Clang 4.0.1 (tags/RELEASE_401/final)] :: Anaconda, Inc. on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import slycot
>>> slycot.__version__
'0.3.4.23.git2894c98'

@murrayrm

@murrayrm great, thank you. 0.3.5 is tagged - could you please publish to PyPI?

Uploaded. SHA256 has is cad98d5ea4f0a034cf398c39189f587620a0b03f1d4b71e77cd622a327f13adf

Navigation Menu

Repository navigation

Release 0.3.5 #70

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Release 0.3.5 #70

Description

Activity

bnavigator commented on Jun 21, 2019

moorepants commented on Jun 21, 2019

roryyorke commented on Jun 22, 2019

roryyorke commented on Jun 22, 2019

moorepants commented on Jun 22, 2019

moorepants commented on Jun 22, 2019

roryyorke commented on Jun 23, 2019

moorepants commented on Jun 23, 2019

murrayrm commented on Jun 23, 2019

roryyorke commented on Jun 23, 2019

murrayrm commented on Jun 23, 2019

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions