Re: Switching Fedora mingw compiler to dwarf2 exceptions

Hi, Personally I haven't got that much need of throwing exceptions across DLLs (my cross-compiling needs are mostly plain C, with a little bit of C++). I think the choice to switch is less because of MS VC does dwarf2, but that it is the choice of native mingw gcc (running on windows), and if fedora doesn't switch, there would be mix-and-match problems using DLLs built by native mingw gcc. So it is less about being compatible with MSVC but being compatible with native mingw on windows. Most people *on windows* who needs a *free compiler* would just use the supplied mingw/cygwin gcc binaries. FWIW, I have opted for an have-my-cake and eat it approach: http://sourceforge.net/projects/outmodedbonsai/files/Mingw Cross-compiler/ I used to have my own All-in-one cross-compiler rpm under /opt/mingw during F9/F10, and till the middle of F11 (and was taking F12 mingw-* rpm's and rebuilding them under F11) so I could have them side-by-side, at two different locations for a few months during the transition. I have the usual fedora cross-compiling set-up under /usr/bin, but if I choose to set PATH=/opt/mingw/bin:${PATH} , I get "the other" set up. For people who are undecided, this might be a viable approach for a while. Sorry the All-in-One rpm listing only have the SPEC file and the src patch bundle (i.e. without source tar balls that you can download from elsewhere). I never build the src rpm, (it is mostly just sym-links some source tarhall into $RPM_SOURCE_DIR and run rpm -bb ) but I'll answer questions about how they were built as far as I can. The fedora-style mingw-gcc packages are also a little ahead of F12's - I take the native gcc src rpm and rip the contents (snapshot tarballs +fedora patches). The all-in-one rpm follows the mingw conventions - use gcc release tar balls (not snapshots) and the mignw patches, without the fedora patches. Hin-Tak --- On Mon, 30/11/09, Kalev Lember <kalev(a)smartlink.ee&gt; wrote: