This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: a .NET alternative (GJC et al)


Fergus Henderson <fjh@cs.mu.oz.au> writes:

> Certainly it could be done without support from the GCC back-end.

Are you sure? Certain types of optimizations could result in code
which drops a pointer to a specific object by some arithmetic on it
and later resurrects it.

In addition, isn't the determination of the exact root set only
possible with type-accurate stack scanning which seems to imply
backend support?  Issuing code to register pointer sites doesn't seem
to be a reasonable approach to me.

> For most of us the main reason to use type-accurate GC rather than
> conservative GC is to get better performance.  An approach like that
> would defeat the purpose.

Are you sure?  I feel somewhat strange when openly trading correctness
for speed. ;-)

-- 
Florian Weimer 	                  Florian.Weimer@RUS.Uni-Stuttgart.DE
University of Stuttgart           http://cert.uni-stuttgart.de/
RUS-CERT                          +49-711-685-5973/fax +49-711-685-5898


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]