[SYSTEMDS-3097] Tutorial simple-aggregation-example does not work

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: SystemDS 2.1
• Fix Version/s: SystemDS 2.1
• Component/s: Documentation
• Labels: documentation pull-request-available
• Assignee: Unassigned
• Reporter: Martin
• Created: 17/Aug/21 13:37
• Updated: 23/Aug/21 10:36
• Resolved: 23/Aug/21 09:51

Description

I am using systemds, commit 59267c983bc7632f7c779618f4824114dd6ff4e4. I tried to run the "Simple Aggregation Example" tutorial from docs/source/guide/federated.rst or from https://apache.github.io/systemds/api/python/guide/federated.html#simple-aggregation-example but it failed.

Traceback (most recent call last):
[...]
fed_a = sds.federated(sds, [address], [dims])
File "/home/martin/anaconda3/envs/systemds/lib/python3.9/site-packages/systemds-2.2.0-py3.9.egg/systemds/context/systemds_context.py", line 440, in federated
','.join(map(lambda s: f'"{s}"', addresses)) + ')'
TypeError: 'SystemDSContext' object is not iterable

 

I applied following change:

fed_a = sds.federated(sds, [address], [dims])

fed_a = sds.federated([address], [dims])

 

Now, the script runs through and shows the result "39". This is not the expected value of 45. It seems the first row of "temp/test.csv" is not included. As a quick-fix, one could add an empty first row to the csv-file:

np.savetxt("temp/test.csv", a, delimiter=",")

np.savetxt("temp/test.csv", a, delimiter=",", header=' ')

Issue Links

• links to: GitHub Pull Request #1366
• links to: GitHub Pull Request #1370