Uh oh!
There was an error while loading. Please reload this page.
ZEPPELIN-346 Fix NotebookTest not passing non-default group to NoteIn… - #347
ZEPPELIN-346 Fix NotebookTest not passing non-default group to NoteIn…#347RPCMoritz wants to merge 1 commit into
Conversation
…terpreterLoader This implements the actually required logic for this test to pass and should mostly showcase that the NoteIntepreterLoader logic is flawed/broken. Nonetheless this would fix the test in the short term and be a step towards actually documenting the weird behaviour. I will try to amend this PR (or create a separate one) to actually approach the underlying issue
RPCMoritz
commented
Oct 30, 2015
@Leemoonsoo if you plan to release a new version, then you should probably merge this as-is. |
Leemoonsoo
commented
Oct 30, 2015
@RPCMoritz Thanks for making the patch and interest to Zeppelin. Selecting interpreter using %[group].[name] notation introduced by https://issues.apache.org/jira/browse/ZEPPELIN-74. Main reason the logic working in this way is to keep compatibility with previous behavior, ie. to keep selecting interpreter simple. So, I'd like to understand more about why do you think there're bad test and ambiguous logic, and what are you trying to fix in the end. Could you explain more? |
RPCMoritz
commented
Oct 30, 2015
I believe the test currently only passes CI, because of the broken implementation of using System properties. Without setting group.name explicetely, the guessing code does not correctly guess the second interpreter by name only, because the structure used does not match how it's written into system properties (in this test). The actual issue may be in the module writing system properties, but I assert that using System properties in this way is not ideal, and using API calls or a dedicated configuration object instead is more useful (but still not generally independent across multiple notebooks and therefore influenced by tests being run in parallel) So the current fix to the test makes tests pass in linear execution. I looked at this issue together with @FRosner , who gave his commentary in the JIRA issue, which may be helpful in making my point. |
Leemoonsoo
commented
Oct 30, 2015
Could you help me to reproduce the test error? Because of the test also passes in my machines, both using mvn test command and run inside of IDE. |
RPCMoritz
commented
Oct 30, 2015
On any machine I attempted to build, I got the test failure (interpreter not found). I'll try again with the current master on yet another machine to see if I can still reproduce. |
FRosner
commented
Oct 30, 2015
@Leemoonsoo the test also failed on my machine at the time the patch was proposed. I still think, however, that passing two parameters as one by splitting a String is a very bad idea. See https://issues.apache.org/jira/browse/ZEPPELIN-346?focusedCommentId=14960634&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14960634 |
bzz
commented
Jan 5, 2016
@RPCMoritz@FRosner does the issue still exist on the latest master? |
RPCMoritz
commented
Jan 5, 2016
I will re-test on my systems ASAP. |
RPCMoritz
commented
Jan 8, 2016
testSelectingReplImplementation(org.apache.zeppelin.notebook.NotebookTest) Time elapsed: 0.032 sec <<< ERROR! with Apache Maven 3.2.5 (12a6b3acb947671f09b81f49094c53f426d8cea1; 2014-12-14T18:29:23+01:00) So, still broken. |
Add Tutorial Notebook
close#83close#86close#125close#133close#139close#146close#193close#203close#246close#262close#264close#273close#291close#299close#320close#347close#389close#413close#423close#543close#560close#658close#670close#728close#765close#777close#782close#783close#812close#822close#841close#843close#878close#884close#918close#989close#1076close#1135close#1187close#1231close#1304close#1316close#1361close#1385close#1390close#1414close#1422close#1425close#1447close#1458close#1466close#1485close#1492close#1495close#1497close#1536close#1545close#1561close#1577close#1600close#1603close#1678close#1695close#1739close#1748close#1765close#1767close#1776close#1783close#1799
…terpreterLoader
This implements the actually required logic for this test to pass and should mostly showcase that the NoteIntepreterLoader logic is flawed/broken. Nonetheless this would fix the test in the short term and be a step towards actually documenting the weird behaviour.
I will try to amend this PR (or create a separate one) to actually approach the underlying issue