hg: jdk7/tl/langtools: 6917130: should test that annotations that have been optimized away are not emitted to classfile

[Jonathan Gibbons]

Neal,

That may be so, but that is not currently a goal for the langtools 
regression test suite, nor do we currently have any way of marking tests 
as compiler-specific or compiler-independent.

-- Jon


Neal Gafter wrote:
> The javac regression suite has been very helpful to me in finding bugs 
> in other compilers that were not caught by those other suites.  That's 
> because many of the javac regression tests exercise delicate 
> interactions between language features.  That's quite valuable to the 
> Java ecosystem.  But using the suite that way requires separating the 
> compiler-specific and compiler-independent parts of the suite.  It's 
> far easier to do that in one central place - the suite itself - than 
> separately by everyone who benefits from using it that way.
>
> On Tue, Jan 26, 2010 at 12:46 PM, Joseph D. Darcy <Joe.Darcy at sun.com 
> <mailto:Joe.Darcy at sun.com>> wrote:
>
>     Jonathan Gibbons wrote:
>
>         Neal Gafter wrote:
>
>             On Tue, Jan 26, 2010 at 12:24 PM, Joseph D. Darcy
>             <Joe.Darcy at sun.com <mailto:Joe.Darcy at sun.com>> wrote:
>              
>
>                 Neal Gafter wrote:
>                    
>
>                     Note that this is testing for "optional" behavior,
>                     which is a
>                     quality-of-implementation issue and not a
>                     correctness issue.
>                          
>
>                 ...which is perfectly appropriate for JDK-specific
>                 regression tests :-)
>                    
>
>
>             Agreed.  However, it would be a good idea to put a note in
>             the test,
>             just in case someone ends up confused over why some other
>             compiler
>             doesn't pass it.
>              
>
>
>         Well, there is no precedent for making any such notes in the
>         many other tests for
>         javac-specific tests. Look at all the tests that directly use
>         internal javac API.
>
>
>     All the tests in the langtools repository come with the implicit
>     requirement of being run against the matching javac.
>
>     I think putting explicit notes on this point in particular tests
>     is generally unnecessary and would be undesirable.
>
>     There are other compiler and language test suites explicitly
>     designed and intended to be compiler agnostic.
>
>     -Joe
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/compiler-dev/attachments/20100126/7709ddfd/attachment.html