Eli Zaretskii
eliz****@gnu*****
Sat Jul 28 22:03:24 JST 2018
> From: Keith Marshall <keith****@users*****> > Date: Tue, 3 Jul 2018 15:39:56 +0100 > > I've had a change of heart; or rather, while comparing David's build > specification with my own, I had one of those "light-bulb" moments. > (Specifically, GCC's configure scripts don't respect the convention, as > normal autoconf generated scripts *should*, that $target may be inferred > by propagation from $host, unless explicitly specified otherwise). > > When I explicitly specify "--target=mingw32", in addition to the already > required "--host=mingw32", then I can achieve a successful build; > however, the step at which the build previously choked -- a self-test > feature which is new in gcc-7.x -- still seems to be of questionable > value, for it tests the capabilities of the build compiler, rather than > the one we're actually building, (contrary to the intent indicated by > comments in the Makefile)! > > >> You might want to try building it yourself; My Windows native build > >> has very few problems. > > > > I might do that if I find enough free time. > > Or, you may prefer to try the binaries, which I've now published at: > https://osdn.net/projects/mingw/releases/p15691 A belated thank you: I have no installed GCC 7.3.0 on my main development machine, and all the tests so far pass with flying colors. Thank you!
Fork