This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: Symbol table resolution
- From: Bryce McKinlay <mckinlay at redhat dot com>
- To: Andrew Haley <aph at redhat dot com>
- Cc: java at gcc dot gnu dot org
- Date: Wed, 07 Jun 2006 14:20:26 -0400
- Subject: Re: Symbol table resolution
- References: <17543.3948.293570.795837@dell.pink>
Andrew Haley wrote:
Properly enforcing access rules in libgcj causes problems in that
compiler-generated code sometimes wants to access private methods.
Presumably, this is only because GCJ doesn't generate accessor methods
correctly? [*]
If so, the issue should go away when we switch to the ecj front-end. Is
it worth putting in this hack now for something that really shouldn't
occur?
Also, won't this still break existing binaries? We'd need to bump the
ABI version and only enable the check for new binaries.
If we want to keep the direct private calls & field accesses as an ABI
optimization, then we could instead check that the class being accessed
is a nested or enclosing class (ie, check for compliance with the JLS
rules rather than the JVMS rules).
Bryce
[*] http://gcc.gnu.org/PR19870