[ https://issues.apache.org/jira/browse/IGNITE-4211?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15937782#comment-15937782 ]
Vyacheslav Daradur edited comment on IGNITE-4211 at 3/23/17 6:04 AM: --------------------------------------------------------------------- [~avinogradov] I knew about spring-aop-methods before, but I couldn't understand what you want. bq. What? As I've written here before, I think we have to cover the all new code with unit tests (if possible and where necessary). I've added the new get-method. public <T> T get(Object key, Callable<T> valueLoader) And I think we have to tests the following cases: - valueLoader parameter was called only once per same key; - if a cache contains nothing for a key, a result of valueLoader parameter will added to the cache correctly; - if valueLoader parameter throws some exceptions, the new get-method rethrows the excepted type of exception; These are main cases wich described in the java-docs on the new method. All cases above were checked in the unit-tests which you don't like. bq. Why? Just make that the any get-methods returned "null". What happens? - Nothing. Green light in the GridSpringCacheManagerSelfTest. Tests in the GridSpringCacheManagerSelfTest don't check correct behavior of get-methods. Maybe I'm wrong and returned value are checked in the other tests. (other cases of behavior?) If you don't agree with my approach, and if the test of @Cachenable(sync=true) is enough for you: Just write it, I will change the test and then it's done. Such simple task was delayed for a long time. was (Author: daradurvs): [~avinogradov] I knew about spring-aop-methods before, but I couldn't understand what you want. bq. What? As I've written here before, I think we have to cover with the all new code unit tests, if possible and where necessary. I've added the new get-method. public <T> T get(Object key, Callable<T> valueLoader) And I think we have to tests the following cases: - valueLoader parameter calls only once per same key; - if cache contain nothing for key, result of valueLoader parameter will added to cache correctly; - if valueLoader parameter throws some exceptions, new get-method rethrow the excepted type of exception; These are main cases wich described in the java-docs on the new method. All cases above were checked in the unit-tests which you don't like. bq. Why? Just make that the any get-methods returned "null". What happens? - Nothing. Green light in the GridSpringCacheManagerSelfTest. Tests in the GridSpringCacheManagerSelfTest don't check correct behavior of get-methods. Maybe I'm wrong and returned value are checked in the other tests. (other cases of behavior?) If you don't agree with my approach, and if the test of @Cachenable(sync=true) is enough for you: Just write it, I will change the test and then it's done. Such simple task was delayed for a long time. > Update Spring dependency to latest stable version > ------------------------------------------------- > > Key: IGNITE-4211 > URL: https://issues.apache.org/jira/browse/IGNITE-4211 > Project: Ignite > Issue Type: Improvement > Components: build > Affects Versions: 1.7 > Reporter: Sergey Kozlov > Assignee: Vyacheslav Daradur > Fix For: 2.0 > > > It seems the Spring dependency looks outdated for now. Apache Ignite still > uses 4.1.0 released two years ago. Could we to update to latest stable > version (4.3.4 at the moment)? -- This message was sent by Atlassian JIRA (v6.3.15#6346)