Mail Archives: djgpp/1999/01/29/03:16:46
On 28 Jan 1999, Mumit Khan wrote:
> In article <Pine DOT SUN DOT 3 DOT 91 DOT 990128162725 DOT 29318B-100000 AT is>,
> Eli Zaretskii <djgpp AT delorie DOT com> wrote:
> >
> >On Thu, 28 Jan 1999, Paul Derbyshire wrote:
> >
> >> In other words, C++ support is basically exactly where it was a year ago :P
> >
> >You can't be serious. A year ago you didn't have exceptions supported at
> >all, IIRC, and templates had serious bugs. GCC 2.8.1 fixed these
> >problems, AFAIK.
>
> Actually, that's not quite accurate. Both templates and exception handling
> are quite broken in GCC-2.8.1, and remain so. There is currently *NO* C++
> development ongoing in the FSF GCC distribution.
>
> If exception handling doesn't work with djgpp port of EGCS due to bugs in
> EGCS, then the issue should be reported to egcs folks (is there a djgpp
> egcs maintainer).
>
> If djgpp egcs port suffers from EH problems and is using dwarf2 unwind,
> perhaps you ought to try out setjmp/longjmp based EH and see if that
> works. This is essentially what we ended up doing for all the win32
> ports.
>
Actual problems why C++ exceptions doesn't work reliably with DJGPP port
of egcs is found (Lazlo Molnar and I). There is no need to use
-fsjlj-exceptions. Seems that after corresponding fixes there is no
more significant exceptions related regressions against gcc-2.8.1 at all
or egcs-1.1.1 under Linux.
About Win32 ports: I may be wrong but anyway I think it's worth to try -
download some recent binutils snapshot and build it with
configure option --enable-bfd-assembler and see if it works.
At least I can say it helps for DJGPP port of egcs-1.1.1
(I and Lazlo have tested that with AS.EXE I have built from
990120 binutils snapshot)
Andris
- Raw text -