finds classes in precompiled folder too

[cbxp]

currently Play 1.5 cannot find any Entity classes in precompiled mode

[flybyray]

wait a minute, that patch makes no sense. in precompiled mode all classes are loaded before any plugin like the JPAPlugin is doing real work. if your entities are not there there is some other problem.
that fix may resolve somewhat the symptom but not the real problem. but just to late for today, i try to investigate that another time. at least it gave a hint, because those changes in applicationclassloader for the hibernate5 update is suspicious.

[asolntsev]

Yes, all class are loaded. But look what method has been changed: it's "getResource()", not "getClass()". Unfortunately Hibernate5 call method ClassLoader.getResource("com/my/SomeEntity.class") just to detect if entity class exists.

…

On May 27, 2017 2:25 AM, "Robert Rettig" ***@***.***> wrote: wait a minute, that patch makes no sense. in precompiled mode all classes are loaded before any plugin like the JPAPlugin is doing real work. if your entities are not there there is some other problem. that fix may resolve somewhat the symptom but not the real problem. but just to late for today, i try to investigate that another time. at least it gave a hint, because those changes in applicationclassloader for the hibernate5 update is suspicious. — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#1152 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AARE3TtEEGjZTyQw88feR-XG1vO0eNGXks5r919YgaJpZM4NnfcO> .

[flybyray]

just glad to have seen #1152
somehow it is good to describe errors in detail (stacktrace) on lighthouse or an github issue first.
then someone can make some reliable decission how to fix the real problem.
a classloader change just for upgrading a library is suspicious i did not recognized it by a fast review of #1114 .
i think the real fix could be something like #1155

[asolntsev]

I like your change. It's really simpler and better solution for loading entity classes. Thought, my classloader change is not bad. It's a common practice that method ClassLoader.getResource() can return *.class files. Most classloaders do it.

…

On May 29, 2017 7:48 AM, "Robert Rettig" ***@***.***> wrote: just glad to have seen #1152 <#1152> somehow it is good to describe errors in detail (stacktrace) on lighthouse or an github issue first. then someone can make some reliable decission how to fix the real problem. reviewing #1141 <#1141> was to fast, an classloader change just for upgrading a library is suspicious. i think the real fix could be something like #1155 <#1155> — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#1152 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AARE3XB8NMYdR8asfFm9M2BJ36HhRACpks5r-k4tgaJpZM4NnfcO> .

[flybyray]

yes might be a good change in some (odd) cases. it was not needed before, maybe it would be interessting to have a trace logging in those conditional branches to see if it would effect any test case in play. maybe it should stay a feature which is turned off by default.
but in general separation of concerns should be applied to pull requests too. a library upgrade should result in minimum changes to existing source.

ps: those many pull request from my side lastly is a result of this https://play.lighthouseapp.com/projects/57987-play-framework/tickets/2067-fix-os-and-language-related-stuff#ticket-2067-5

ps: hmm a spam bot? https://play.lighthouseapp.com/projects/57987-play-framework/tickets?q=created%3A04-26-17&filter=

[flybyray]

@asolntsev

It's really simpler and better solution for loading entity classes.

maybe not. as i stated in #1114 (review)
there is some commented todo in hibernate. the loaded_classes is for internal use . but it is for now the only way to get the behavior as it was with 4.x hibernate