Closed fmatthies closed 7 months ago
Some commits are not properly rebased from master
- e.g. f124dff (has it been cherry-picked?)
When the tests are passing, I'll drop the affected commits and do a rebase afterwards.
I did indeed wonder why there were only pom
changes when I was rebasing. I didn't intentionally omit commits... don't know why this happened
Something is off with the application properties file:
java.lang.IllegalArgumentException: Could not resolve placeholder 'top.documents.document_server.batch_size' in value "${top.documents.document_server.batch_size}"
This is weird (wrote about it in the last commit message). Need to look into it.
And it just happens during tests because locally my backend runs with the new changes
And it just happens during tests because locally my backend runs with the new changes
It looks to me like you are modifying the application context in TopBackendApplication::main
, which is not called when the tests are run.
ApplicationContextInitializer should do the trick.
Yeah, you're right. But it never happened before (maybe I have defaults in the value annotion everywhere else); that threw me off.
It's so frustrating: locally the test are passing. I will look into it some more. Hopefully you aren't spammed with the notifications about the failing tests?
@ChristophB So the real reason for all the failures (I fixed them now except for the getDataSources
Test in QueryApiDelegateImplTest
) seems to be that the test/resources
are not used (or old? ones). So the getDataSources
Test just finds 2 (and not my nlp/Test_Data_Source_3.yml
). Maybe you have some insights?
seems to be that the test/resources are not used (or old? ones)
That is not the case. The files there are still used for the tests and the path must be configured in src/test/resources/application.yml
. The value for top.documents.data-source-config-dir
is wrong and leads to the failing test.
I updated the value in the most recent commit, but this leads to failing document service tests. Looks like something is off with the application context rewrite you have implemented.
Now that the application finds an adapter, it will use it to set the default values right?
Not default ones but the ones declared in the adapter. The remaining error resulted from the test documents being indexed not in test_documents
but in documents
. Hopefully my commit fixes that.
Thanks for your help! Don't know how long it would have taken me to arrive at that solution 🤔
Some commits are not properly rebased from
master
- e.g. f124dff (has it been cherry-picked?)When the tests are passing, I'll drop the affected commits and do a rebase afterwards.
@ChristophB now all tests are passing; finally.
Cleaned up the branch and rebased onto master.
Done. (I'll update the readme
in top-document-query
accordingly).
Shall I squash and close or do you want to do it?
I'll do the merge as there is another unwanted commit 005409b.
@fmatthies What is the purpose of this commit? abb59ae It is from January 18 and introduced an unused method.
EDIT: The changes are already included in a previous commit. So I'll just drop that commit.
there are now methods to start the concept graph creation using an external document server and not relying on txt document upload