[jira] Updated: (THRIFT-1019) java backend should use fully qualified names for domain types

["Bryan Duxbury (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/THRIFT-1019?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Bryan Duxbury updated THRIFT-1019:
----------------------------------

    Fix Version/s: 0.7

> java backend should use fully qualified names for domain types
> --------------------------------------------------------------
>
>                 Key: THRIFT-1019
>                 URL: https://issues.apache.org/jira/browse/THRIFT-1019
>             Project: Thrift
>          Issue Type: Bug
>          Components: Java - Compiler
>            Reporter: Peter Schuller
>            Priority: Minor
>             Fix For: 0.7
>
>
> This is a follow-up to THRIFT-1013. The left-overs from that is to also do fully qualified names for the domain classes (for example even with the changes in THRIFT-1013, our 'State' still conflicted because an inner class used for asynch method calls extended the TAsynchMethodCall, bringing State into scope).
> Simply changing type_name() to always generate fully qualified names (not just when program differs) was very simple, but unfortunately was not enough. This exposes the issues that inner classes are not accessible by "package.classname" but rather as "package.outerclass.classname".
> I found no trivial way to fix this because it requires maintaining some new state that is specific to the Java backend, and it affects code (particularly the client bit) that I'm not in a good position to test (currently using it only for server-side). I'm having to drop this for time reasons and instead rename the structure in our thrift spec for now, but I wanted to file the issue anyway even though i don't expect to submit a patch.
> Briefly, what I would have started doing (though I'm not comfortable saying it's the cleanest way to do it), was to add a map member to the generator where one would put the information that maps a t_struct to its enclosing t_struct. This could be filled in while looping in generate_service_helpers(), and looked at in whenever type_name() is used to refer to a type. I'm not sure whether this is enough or not since I haven't tried it.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.