search

hbwf / mybatis

Automatically exported from code.google.com/p/mybatis
0 stars 0 forks source link

StackOverflowError #782

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
Hi,

I found a weird problem with mybatis, spring and declaring statements with 
databaseId that end up in a StackOverflowError (infinite loop).

mybatis version: 3.1.1
mybatis-spring version: 1.1.1

See the attached project with 2 unit tests. One that fails to load the mybatis 
configuration, and the second one that force the databaseId to null and the 
mybatis configuration can be loaded.

I've done a bit of debugging to try and understand the problem and here is what 
I think is the problem:

(Refer to the attached project for files and content)

 - Mapping A is loaded
   - XMLMapperBuilder.buildStatementFromContext is executed with requiredDatabaseId = h2
     - 3 statements (delete + updates) without databaseId are skipped
     - same 3 statements (delete + updates) with databaseId=h2 are loaded
     - select without databaseId is skipped
   - XMLMapperBuilder.buildStatementFromContext is executed with requiredDatabaseId = null
     - 3 statements (delete + updates) with databaseId=h2 are skipped
     - same 3 statements (delete + updates) without databaseId are skipped because "skip this statement if there is a previous one with a not null databaseId" (databaseIdMatchesCurrent method)
     - select without databaseId is added to the incomplete statements because it includes sql from mapper C

 - Mapping B is loaded, similar to A except:
   - XMLMapperBuilder.buildStatementFromContext is executed with requiredDatabaseId = null
     - 3 statements (delete + updates) without databaseId are parsed, they should be skipped but the following happens instead
       - in databaseIdMatchesCurrent
         - configuration.hasStatement is called with validateIncompleteStatements set to false and returns true
         - configuration.getMappedStatement with default value for validateIncompleteStatements set to true
           - it tries to parse the statement from mapper A (still incomplete) and throws an exception
           - The statement is added to the incomplete statements

 - Mapping C is loaded
   - Again in databaseIdMatchesCurrent it tries to resolve incomplete statements, but this time mybatis knows about mapping C, so when processing the select from mapping A, it also enters into the databaseIdMatchesCurrent which tries to resolve incomplete statements, loads select from A again, and gets into an infinite loop.

The good news is that I think this is a very simple fix, 
XMLStatementBuilder.databaseIdMatchesCurrent line 130 should set 
validateIncompleteStatements to false: "MappedStatement previous = 
this.configuration.getMappedStatement(id, false);"

Also what is strange is I could not reproduce the problem without loading 
mybatis through spring, but I don't see why that would matter.

Hope the fix is as easy as that and it can be fixed for the next version.

My current workaround is to load the mapping in a different order (B, C and 
then A).
But there might be a time soon where all our mappings will reference sql 
fragments from other files and we won't be able to use that workaround.

Thanks

Original issue reported on code.google.com by dcende...@gmail.com on 21 Feb 2013 at 2:19

Attachments:

GoogleCodeExporter commented 9 years ago

Original comment by eduardo.macarron on 24 Feb 2013 at 7:42

GoogleCodeExporter commented 9 years ago 
You are right, validateIncompleteStatements is meant to be used when 
initialization is done, not during startup. So that looks like the right fix.

Thank you!

The snapshot is updated. Could you please confirm it is fixed?

Original comment by eduardo.macarron on 24 Feb 2013 at 9:30

GoogleCodeExporter commented 9 years ago

Original comment by eduardo.macarron on 28 Feb 2013 at 6:02

GoogleCodeExporter commented 9 years ago 

Hi,

I've tested the latest code from the master branch and it does fix the problem.

BTW, you should update the website to let the users know the source code has 
been moved to github, I first tried using the subversion repository and 
obviously it didn't work.

The following page still shows subversion as the main repository: 
http://code.google.com/p/mybatis/wiki/Contribute

Cheers

Original comment by dcende...@gmail.com on 28 Feb 2013 at 11:26

GoogleCodeExporter commented 9 years ago 
Hi!

Thanks for the testing. 

We will annouce the movement officially soon. Sorry for the misleading 
information.

Original comment by eduardo.macarron on 6 Mar 2013 at 3:13