This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: A completely different approach to EH runtime
- To: David Edelsohn <dje at watson dot ibm dot com>
- Subject: Re: A completely different approach to EH runtime
- From: "Zack Weinberg" <zackw at Stanford dot EDU>
- Date: Wed, 21 Feb 2001 11:15:15 -0800
- Cc: Hidvegi <hzoli at austin dot ibm dot com>, gcc at gcc dot gnu dot org
- References: <hzoli@austin.ibm.com> <200102211757.MAA29608@mal-ach.watson.ibm.com>
On Wed, Feb 21, 2001 at 12:57:53PM -0500, David Edelsohn wrote:
> >>>>> Hidvegi writes:
>
> Hidvegi> If I understand this correctly, this would mean that you cannot load a
> Hidvegi> g++ compiled module dynamically from a program that was not compiled
> Hidvegi> with gcc.
>
> Yes. This is a serious problem with Zack's proposal. It means
> that G++-produced shared objects could not be loaded through Java JNI, for
> example.
That does look like a show-stopper. Drat.
Hm, the whole problem is that there has to be exactly one copy of the
global data in _eh.o and frame.o. Could we bodge up something with
unique weak symbols so that (at least on some platforms) there'd be
only one copy of the global frame data no matter how many times the
library got dragged in?
zw