On Wed, 2007-12-26 at 10:46 +0000, Pedro Alves wrote:
> Danny Backx wrote:
> > As I said I had the complete build working on mingw32ce. But not
> > everything was working yet, and after I did my changes,
> 
> > And it appears that I didn't do a commit in all dirs yet (w32api and
> > mingw were still missing).
> >
> 
> Let's try to keep the trunk building at all times, OK ?
> If the change was incomplete, then you shouldn't have changed the
> default target name in the build script.

That was the idea, except I found out too late that I'd forgotten to
commit some of the stuff. The build was working, and a whole bunch of
test programs compiled, linked, and worked just fine.

> Did I mention more than once already that I'd like to follow that Danny Smith
> has done with the latest MinGW gcc4.2.1 release?

> "* To throw exception across dll/exe boundaries in C++, you will need to
> link against the shared libgcc and libstdc++ runtime libs. To link
> against shared libgcc, add -shared-libgcc to command line. To link
> against shared libstdc++ add -lstdc++_s to command line. Hopefully, in
> future releases, the mechanism to seletc static or shared runtime libs
> can be improves. Unless you want to use auto-import to import data
> objects in libstdc++.dll, also add -D_DLL to commands line or #define
> _DLL in top-level headers. Again, this may chenage in future releases."

Not with this amount of detail, at least not to my knowledge.

> It is important for us to follow the same options MinGW uses.
> 
> 1) Leverage their knowledge, understand their work.  Our script was copied
>     Danny already, but he has something new now.

This is certainly new to me.

> 2) Consistency with the big brother.  In the future when those options start
>     showing in MinGW ports of stuff, we'd have 0 porting effort.
> 3) Leave static runtime as default, so we don't pull the rug under our
>     users feet.

I don't agree on all the assumptions here. In an earlier message you
replied to my question about use of build-mingw32ce-dll.sh that it was
insufficiently tested.

My motivations would be :
a. Get something working, get it out to the world so people can use it.
   Hence my move *after* a release to start using build-mingw32ce-dll.sh
b. Try to minimize porting effort for yourself and for the people who
   use it. (I'd like the effort to be 0 but I'm not counting on it.)
c. Provide a way for people who are wary of experimental stuff to avoid
   it so they can choose for stability at all times.

I think your list and mine conflict "somewhat". Can we find a way ahead
that is "good" for both of us ?

        Danny

Attachment: signature.asc
Description: This is a digitally signed message part

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Cegcc-devel mailing list
Cegcc-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/cegcc-devel

Reply via email to