Re: Bytecode verifier

[Ranjit Mathew <rmathew at hotmail dot com>]

Tom Tromey wrote:
> Ranjit> If libgcj provides the API for the verifier and the
> Ranjit> front-end uses it, the configury and the build process
> Ranjit> should get really weird.
> 
> At the moment the API is just a single .h file declaring all the
> functions and types used by the verifier that are provided by the
> enclosing environment.  We can put the libgcj version of this file
> under the same copyright as the rest of libgcj.  There's really no
> problem here.

Cool! IMHO, you should then really update and apply your
patch to the mainline, if possible, so that we do
not have to maintain two verifiers.


> Ranjit> not to bother by default. A runtime flag/environment-
> Ranjit> variable/property should at least help in not having
> Ranjit> to comment out the verifier call and recompiling
> Ranjit> libgcj in order to get an application working.
> 
> Yeah.  Could you file this in bugzilla as a feature request?
> I thought we already had it in there, but I don't see it.

Done:

  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14781

Ranjit.

-- 
Ranjit Mathew          Email: rmathew AT hotmail DOT com

Bangalore, INDIA.      Web: http://ranjitmathew.tripod.com/