Test failure on CI - Githubissues

ashander commented 7 years ago

eg

    def check_tables(args):
        nodes = args.node_table()
        assert(nodes.num_rows == args.num_nodes)
        edgesets = args.edgeset_table()
        # check edgesets are in order and all parents are recorded
>       node_times = nodes.time
E       AttributeError: 'NodeTable' object has no attribute 'time'

tests/test_merge_records_with_wf.py:11: AttributeError

petrelharp commented 7 years ago

>>> import msprime
>>> node = msprime.NodeTable()
>>> node.time
array([], dtype=float64)
>>> print(msprime.__version__)
0.4.1.dev356+ng85413c9

One commit before the one I said above, but this hasn't changed.

On Thu, Apr 13, 2017 at 9:43 AM, ashander notifications@github.com wrote:

eg
def check_tables(args):
    nodes = args.node_table()
    assert(nodes.num_rows == args.num_nodes)
    edgesets = args.edgeset_table()
    # check edgesets are in order and all parents are recorded
  node_times = nodes.time
E AttributeError: 'NodeTable' object has no attribute 'time'
tests/test_merge_records_with_wf.py:11: AttributeError

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ashander/ftprime/issues/18, or mute the thread https://github.com/notifications/unsubscribe-auth/AA_26Y1MSIWF4U8em_5ZqRY0s04PsgiTks5rvlDKgaJpZM4M88K3 .

ashander commented 7 years ago

OK, this works for me too on msprime (0658998) but it's unclear why or how it's not working in CI b/c theenvironment.yml is trying to check out the same commit. :angry:

PS hard to understand these version strings. I don't think they're supposed to be repeatable across compilations, but not sure. in any case, the code below is from the same hash (I've run make just before the commands below)

 $ git log |head -5
commit 06589980d050982390b35f527bb38af9bafd5721
Merge: 85413c9 3d58746
Author: Jerome Kelleher <jk@well.ox.ac.uk>
Date:   Wed Apr 12 18:09:19 2017 +0100

(msprime-dev) jaime@shuksan:~/lib/msprime$ ((0658998...)) 
 $ python
Python 3.6.0 | packaged by conda-forge | (default, Feb  9 2017, 14:36:55) 
[GCC 4.8.2 20140120 (Red Hat 4.8.2-15)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import msprime
>>> node = msprime.NodeTable()
>>> node.time
array([], dtype=float64)
>>> print(msprime.__version__)
0.4.1.dev360+ng0658998

ashander commented 7 years ago

Ok, so can reproduce failure in another env, which is py35. So still unsure, but maybe a py35 vs py36 (as above) issue.

But will look into this tomrrow :\

ashander commented 7 years ago

The underlying issue was my local envs and CI both did not have numpy

ashander commented 7 years ago

Filed https://github.com/jeromekelleher/msprime/issues/189 about this. Seems there should be warning on builiding msprime in an environment without numpy

petrelharp commented 7 years ago

There is - I forget where, but I think just after make ext3.

On Fri, Apr 14, 2017 at 12:50 PM, ashander notifications@github.com wrote:

Filed jeromekelleher/msprime#189 https://github.com/jeromekelleher/msprime/issues/189 about this. Seems there should be warning on builiding msprime in an environment without numpy

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/ashander/ftprime/issues/18#issuecomment-294223570, or mute the thread https://github.com/notifications/unsubscribe-auth/AA_26SY_m4DZV8RBmLBN7-4owo0NRHKnks5rv84ZgaJpZM4M88K3 .

ashander / ftprime

Test failure on CI #18