On Mon, Feb 14, 2011 at 09:21:19AM +0100, Carlos Rivera wrote: > Hi, > > On 12/02/11 00:58, Ron wrote: > > On Fri, Feb 11, 2011 at 01:05:29PM +0100, Carlos Rivera wrote: > > > I have tried with Ubuntu Maverick which has the same upstream tarball > > > (mingw32_4.2.1.dfsg.orig.tar.gz) but a different patch > > > (mingw32_4.2.1.dfsg-2ubuntu1.diff.gz). > > > The compiler works there so I took their patch and diffed it with the > > > ubuntu squeeze patch. > > > > Which part of "this doesn't actually fix the problem" was unclear in the > > initial discussion of this report? > > I did not find "this doesn't actually fix the problem" or similar anywhere in > the report.
The first paragraph describing the bug says: The shared libgcc and libstdc++ runtime libs are missing in mingw32. Without those exceptions can't across dll/exe boundaries. and the first reply to that says: As I understand comments from Danny wrt the 4.2.1 release this isn't the only problem with doing that still. And since nothing has changed to fix that ... > However the headers of the bug report say: > > mingw32: cannot find -lgcc_s > > Fixed in version mingw32/4.2.1.dfsg-1.2 It's not fixed, that's a mess left behind by fabo, who blindly uploaded this same patch without talking to anybody, without understanding the problem, without actually attaching the patch he did apply, and without cleaning up after himself after making busywork for others :( So I'm sorry you wasted your time on a wild-goose chase, it's just a little frustrating to keep seeing patches that hide the symptoms of a problem without actually doing anything at all to really fix it. Doubly so when they are being published by a derived distro which never bothered to get them reviewed, but touts itself as some sort of respected community player whose work you can trust ... > So I figured that if the bug is fixed then one should not get the cannot > find -lgcc_s error when compiling. But then it might just be my ingnorance > of Debian procedures. The symptom you are seeing is that we don't build shared libgcc. Since the only good reason for having that is to enable the exception handling, and since the exception handling is broken ... the problem isn't your grasp of debian procedure so much as a broader trend of people thinking "If I can just make it compile, it will all be ok". In this case, it's not. So by not compiling early, we can save developers a lot of time not trying something that just won't work reliably for them anyway. Is this actually causing trouble for you in real code somewhere? The only 'sure' workaround known at this stage is to go back to the 3.4 release of the toolchain. But possibly it is fixed in mainline gcc now too. If you don't need the exception handling, the static libgcc should work for you just fine and you can ignore this bug completely. Cheers, Ron -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
