[Numpy-discussion] Appveyor Testing Changes

Discussion:

G Young

2016-01-25 22:37:59 UTC

Hello all,

I currently have a branch on my fork (not PR) where I am experimenting with
running Appveyor CI via Virtualenv instead of Conda. I have build running
here <https://ci.appveyor.com/project/gfyoung/numpy-dr74v/build/1.0.14>.
What do people think of using Virtualenv (as we do on Travis) instead of
Conda for testing?

Thanks,

Greg

Nathaniel Smith

2016-01-25 22:50:07 UTC

Permalink

Post by G Young
Hello all,
I currently have a branch on my fork (not PR) where I am experimenting with
running Appveyor CI via Virtualenv instead of Conda. I have build running
here. What do people think of using Virtualenv (as we do on Travis) instead
of Conda for testing?

Can you summarize the advantages and disadvantages that you're aware of?

-n

--
Nathaniel J. Smith -- https://vorpus.org

G Young

2016-01-25 23:21:28 UTC

Permalink

With regards to testing numpy, both Conda and Pip + Virtualenv work quite
well. I have used both to install master and run unit tests, and both pass
with flying colors. This chart here
<http://conda.pydata.org/docs/_downloads/conda-pip-virtualenv-translator.html>
illustrates
my point nicely as well.

However, I can't seem to find / access Conda installations for slightly
older versions of Python (e.g. Python 3.4). Perhaps this is not much of an
issue now with the next release (1.12) being written only for Python 2.7
and Python 3.4 - 5. However, if we were to wind the clock slightly back to
when we were testing 2.6 - 7, 3.2 - 5, I feel Conda falls short in being
able to test on a variety of Python distributions given the nature of Conda
releases. Maybe that situation is no longer the case now, but in the long
term, it could easily happen again.

Greg

Post by G Young

Post by G Young
Hello all,
I currently have a branch on my fork (not PR) where I am experimenting

with

Post by G Young
running Appveyor CI via Virtualenv instead of Conda. I have build

running

Post by G Young
here. What do people think of using Virtualenv (as we do on Travis)

instead

Post by G Young
of Conda for testing?

Can you summarize the advantages and disadvantages that you're aware of?
-n
--
Nathaniel J. Smith -- https://vorpus.org
_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion

Michael Sarahan

2016-01-26 00:08:49 UTC

Permalink

Conda can generally install older versions of python in environments:

conda create -n myenv python=3.4

You really don't need any particular initial version of python/conda in
order to do this. You do, however, need to activate the new environment to
use it:

activate myenv

(For windows, you do not need "source")

Hth,
Michael

Post by G Young
With regards to testing numpy, both Conda and Pip + Virtualenv work quite
well. I have used both to install master and run unit tests, and both pass
with flying colors. This chart here
<http://conda.pydata.org/docs/_downloads/conda-pip-virtualenv-translator.html> illustrates
my point nicely as well.
However, I can't seem to find / access Conda installations for slightly
older versions of Python (e.g. Python 3.4). Perhaps this is not much of an
issue now with the next release (1.12) being written only for Python 2.7
and Python 3.4 - 5. However, if we were to wind the clock slightly back to
when we were testing 2.6 - 7, 3.2 - 5, I feel Conda falls short in being
able to test on a variety of Python distributions given the nature of Conda
releases. Maybe that situation is no longer the case now, but in the long
term, it could easily happen again.
Greg

Post by G Young

Post by G Young
Hello all,
I currently have a branch on my fork (not PR) where I am experimenting

with

Post by G Young
running Appveyor CI via Virtualenv instead of Conda. I have build

running

Post by G Young
here. What do people think of using Virtualenv (as we do on Travis)

instead

Post by G Young
of Conda for testing?

_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion

Bryan Van de Ven

2016-01-26 00:13:25 UTC

Permalink

With regards to testing numpy, both Conda and Pip + Virtualenv work quite well. I have used both to install master and run unit tests, and both pass with flying colors. This chart here illustrates my point nicely as well.
However, I can't seem to find / access Conda installations for slightly older versions of Python (e.g. Python 3.4). Perhaps this is not much of an issue now with the next release (1.12) being written only for Python 2.7 and Python 3.4 - 5. However, if we were to wind the clock slightly back to when we were testing 2.6 - 7, 3.2 - 5, I feel Conda falls short in being able to test on a variety of Python distributions given the nature of Conda releases. Maybe that situation is no longer the case now, but in the long term, it could easily happen again.

Why do you need the installers? The whole point of conda is to be able to create environments with whatever configuration you need. Just pick the newest installer and use "conda create" from there:

***@0199-bryanv (git:streaming) ~/work/bokeh/bokeh $ conda create -n py26 python=2.6
Fetching package metadata: ..............
Solving package specifications: ..........
Package plan for installation in environment /Users/bryan/anaconda/envs/py26:

The following packages will be downloaded:

package | build
---------------------------|-----------------
setuptools-18.0.1 | py26_0 343 KB
pip-7.1.0 | py26_0 1.4 MB
------------------------------------------------------------
Total: 1.7 MB

The following NEW packages will be INSTALLED:

openssl: 1.0.1k-1
pip: 7.1.0-py26_0
python: 2.6.9-1
readline: 6.2-2
setuptools: 18.0.1-py26_0
sqlite: 3.9.2-0
tk: 8.5.18-0
zlib: 1.2.8-0

Proceed ([y]/n)?

G Young

2016-01-26 01:13:32 UTC

Permalink

Ah, yes, that is true. That point had completely escaped my mind. In
light of this, it seems that it's not worth the while then to completely
switch over to pip + virtualenv. It's might be better actually to rewrite
the current Appveyor tests to use environments so that the test suite can
be expanded, though I'm not sure how prudent that is given how slow
Appveyor tests run.

Greg

Post by G Young
With regards to testing numpy, both Conda and Pip + Virtualenv work

quite well. I have used both to install master and run unit tests, and
both pass with flying colors. This chart here illustrates my point nicely
as well.

Post by G Young
However, I can't seem to find / access Conda installations for slightly

older versions of Python (e.g. Python 3.4). Perhaps this is not much of an
issue now with the next release (1.12) being written only for Python 2.7
and Python 3.4 - 5. However, if we were to wind the clock slightly back to
when we were testing 2.6 - 7, 3.2 - 5, I feel Conda falls short in being
able to test on a variety of Python distributions given the nature of Conda
releases. Maybe that situation is no longer the case now, but in the long
term, it could easily happen again.
Why do you need the installers? The whole point of conda is to be able to
create environments with whatever configuration you need. Just pick the
Fetching package metadata: ..............
Solving package specifications: ..........
Package plan for installation in environment
package | build
---------------------------|-----------------
setuptools-18.0.1 | py26_0 343 KB
pip-7.1.0 | py26_0 1.4 MB
------------------------------------------------------------
Total: 1.7 MB
openssl: 1.0.1k-1
pip: 7.1.0-py26_0
python: 2.6.9-1
readline: 6.2-2
setuptools: 18.0.1-py26_0
sqlite: 3.9.2-0
tk: 8.5.18-0
zlib: 1.2.8-0
Proceed ([y]/n)?
_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion

Ralf Gommers

2016-01-26 18:59:36 UTC

Permalink

Post by G Young
Ah, yes, that is true. That point had completely escaped my mind. In
light of this, it seems that it's not worth the while then to completely
switch over to pip + virtualenv. It's might be better actually to rewrite
the current Appveyor tests to use environments so that the test suite can
be expanded, though I'm not sure how prudent that is given how slow
Appveyor tests run.

At the moment Appveyor is already a bit of a bottleneck - it regularly
hasn't started yet when TravisCI is already done. This can be solved via a
paid account, we should seriously consider that when we have a bit more
experience with it (Appveyor tests have been running for less than a month
I think). But it does mean we should go for a sparse test matrix, and use a
more complete one (all Python versions for example) on TravisCI. In the
near future we'll have to add MingwPy test runs to Appveyor. Beyond that
I'm not sure what needs to be added?

Ralf

Post by G Young
Greg

Post by G Young
With regards to testing numpy, both Conda and Pip + Virtualenv work

quite well. I have used both to install master and run unit tests, and
both pass with flying colors. This chart here illustrates my point nicely
as well.

Post by G Young
However, I can't seem to find / access Conda installations for slightly

_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion

G Young

2016-01-26 23:57:18 UTC

Permalink

Perhaps a pip + virtualenv build as well since that's one way that is
mentioned in the online docs for installing source code. I can't think of
anything else beyond that and what you suggested for the time being.

Greg

Post by Ralf Gommers

Post by G Young
Greg

Post by G Young
With regards to testing numpy, both Conda and Pip + Virtualenv work

quite well. I have used both to install master and run unit tests, and
both pass with flying colors. This chart here illustrates my point nicely
as well.

Post by G Young
However, I can't seem to find / access Conda installations for

slightly older versions of Python (e.g. Python 3.4). Perhaps this is not
much of an issue now with the next release (1.12) being written only for
Python 2.7 and Python 3.4 - 5. However, if we were to wind the clock
slightly back to when we were testing 2.6 - 7, 3.2 - 5, I feel Conda falls
short in being able to test on a variety of Python distributions given the
nature of Conda releases. Maybe that situation is no longer the case now,
but in the long term, it could easily happen again.
Why do you need the installers? The whole point of conda is to be able
to create environments with whatever configuration you need. Just pick the
Fetching package metadata: ..............
Solving package specifications: ..........
Package plan for installation in environment
package | build
---------------------------|-----------------
setuptools-18.0.1 | py26_0 343 KB
pip-7.1.0 | py26_0 1.4 MB
------------------------------------------------------------
Total: 1.7 MB
openssl: 1.0.1k-1
pip: 7.1.0-py26_0
python: 2.6.9-1
readline: 6.2-2
setuptools: 18.0.1-py26_0
sqlite: 3.9.2-0
tk: 8.5.18-0
zlib: 1.2.8-0
Proceed ([y]/n)?
_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion

_______________________________________________
NumPy-Discussion mailing list
https://mail.scipy.org/mailman/listinfo/numpy-discussion