hi, i don't really understand the mingw32-runtime-3.15.1-useconds_t.patch. on fedora-9 this patch is required since otherwise got this errors: -------------------------------------------------- i686-pc-mingw32-gcc -c -I/usr/i686-pc-mingw32/sys-root/mingw/include -Wall -fomit-frame-pointer -I. -I./.. -I./../include -nostdinc -iwithprefixbefore include -I ./../../w32api/include usleep.c -o usleep.o usleep.c:29: error: expected ')' before 'useconds' make[1]: *** [usleep.o] Error 1 make[1]: Leaving directory `/tmp/mingwrt-3.15.1-mingw32/mingwex' make: *** [mingwex] Error 1 error: Bad exit status from /var/tmp/rpm-tmp.50915 (%build) -------------------------------------------------- while on rhel/centos-5 with this patch i get this error -------------------------------------------------- i686-pc-mingw32-gcc -c -I/usr/i686-pc-mingw32/sys-root/mingw/include -Wall -fomit-frame-pointer -I. -I./.. -I./../include -nostdinc -iwithprefixbefore include -I ./../../w32api/include usleep.c -o usleep.o usleep.c:32: error: redefinition of typedef 'useconds_t' /usr/i686-pc-mingw32/sys-root/mingw/include/sys/types.h:119: error: previous declaration of 'useconds_t' was here make[1]: *** [usleep.o] Error 1 make[1]: Leaving directory `/home/lfarkas/rpm/BUILD/mingwrt-3.15.1-mingw32/mingwex' make: *** [mingwex] Error 1 error: Bad exit status from /var/tmp/rpm-tmp.38097 (%build) -------------------------------------------------- but without this patch it's compile clean!? also the comment in this patch: /* RWMJ: This is defined in our local sys/types.h but for some reason * isn't picked up. */ is also annoying.
what's more i already build exactly the same packages ie. mingw32-runtime-3.15.1-1 on the same centos-5.2 and build _without_ the above error!??? so the only thing what can i assume that on fedora and the first time on centos i use mingw32-runtime-bootstrap-1-4. which means mingw32-runtime-bootstrap-1-4 not equivalent with mingw32-runtime-3.15.1-1 and imho it's a bug and should have to be fix.
ps. anyway mingw32-runtime-3.14-6 build without this error.
On Tue, Nov 11, 2008 at 12:45:40AM +0100, Farkas Levente wrote:
usleep.c:32: error: redefinition of typedef 'useconds_t' /usr/i686-pc-mingw32/sys-root/mingw/include/sys/types.h:119: error: previous declaration of 'useconds_t' was here
[...]
what's more i already build exactly the same packages ie. mingw32-runtime-3.15.1-1 on the same centos-5.2 and build _without_ the above error!??? so the only thing what can i assume that on fedora and the first time on centos i use mingw32-runtime-bootstrap-1-4. which means mingw32-runtime-bootstrap-1-4 not equivalent with mingw32-runtime-3.15.1-1 and imho it's a bug and should have to be fix.
I don't use the bootstrap packages any more, since all my machines have long been bootstrapped. However it could cause the difference you are seeing, and it may be worth upgrading the bootstrap package to the latest MinGW version to see if you can get it to build. Maybe the patch is redundant with the latest bootstrap package.
Rich.
Richard W.M. Jones wrote:
On Tue, Nov 11, 2008 at 12:45:40AM +0100, Farkas Levente wrote:
usleep.c:32: error: redefinition of typedef 'useconds_t' /usr/i686-pc-mingw32/sys-root/mingw/include/sys/types.h:119: error: previous declaration of 'useconds_t' was here
[...]
what's more i already build exactly the same packages ie. mingw32-runtime-3.15.1-1 on the same centos-5.2 and build _without_ the above error!??? so the only thing what can i assume that on fedora and the first time on centos i use mingw32-runtime-bootstrap-1-4. which means mingw32-runtime-bootstrap-1-4 not equivalent with mingw32-runtime-3.15.1-1 and imho it's a bug and should have to be fix.
I don't use the bootstrap packages any more, since all my machines have long been bootstrapped. However it could cause the difference you are seeing, and it may be worth upgrading the bootstrap package to the latest MinGW version to see if you can get it to build. Maybe the patch is redundant with the latest bootstrap package.
not just redundant with this patch it's not compile!
Richard W.M. Jones wrote:
On Tue, Nov 11, 2008 at 12:45:40AM +0100, Farkas Levente wrote:
usleep.c:32: error: redefinition of typedef 'useconds_t' /usr/i686-pc-mingw32/sys-root/mingw/include/sys/types.h:119: error: previous declaration of 'useconds_t' was here
[...]
what's more i already build exactly the same packages ie. mingw32-runtime-3.15.1-1 on the same centos-5.2 and build _without_ the above error!??? so the only thing what can i assume that on fedora and the first time on centos i use mingw32-runtime-bootstrap-1-4. which means mingw32-runtime-bootstrap-1-4 not equivalent with mingw32-runtime-3.15.1-1 and imho it's a bug and should have to be fix.
I don't use the bootstrap packages any more, since all my machines have long been bootstrapped. However it could cause the difference you are seeing, and it may be worth upgrading the bootstrap package to the latest MinGW version to see if you can get it to build. Maybe the patch is redundant with the latest bootstrap package.
and how does it supposed to work in koji? what will be included in mock mingw32-runtime-bootstrap or mingw32-runtime? imho it's always a must that these two packages be in sync.
On Tue, Nov 11, 2008 at 02:57:38PM +0100, Farkas Levente wrote:
Richard W.M. Jones wrote:
I don't use the bootstrap packages any more, since all my machines have long been bootstrapped. However it could cause the difference you are seeing, and it may be worth upgrading the bootstrap package to the latest MinGW version to see if you can get it to build. Maybe the patch is redundant with the latest bootstrap package.
and how does it supposed to work in koji? what will be included in mock mingw32-runtime-bootstrap or mingw32-runtime? imho it's always a must that these two packages be in sync.
Well it'll only be an issue for Koji on the very first time that we build those packages. If it is causing a problem, please suggest a patch for the specfile for the *-bootstrap packages, otherwise we'll just sort it out later when we get rel-eng to bootstrap koji.
Rich.
Richard W.M. Jones wrote:
On Tue, Nov 11, 2008 at 02:57:38PM +0100, Farkas Levente wrote:
Richard W.M. Jones wrote:
I don't use the bootstrap packages any more, since all my machines have long been bootstrapped. However it could cause the difference you are seeing, and it may be worth upgrading the bootstrap package to the latest MinGW version to see if you can get it to build. Maybe the patch is redundant with the latest bootstrap package.
and how does it supposed to work in koji? what will be included in mock mingw32-runtime-bootstrap or mingw32-runtime? imho it's always a must that these two packages be in sync.
Well it'll only be an issue for Koji on the very first time that we build those packages. If it is causing a problem, please suggest a patch for the specfile for the *-bootstrap packages, otherwise we'll just sort it out later when we get rel-eng to bootstrap koji.
currently koji able to build any of the mingw32-x packages? i means it's know that he has to use *-bootstrap packages?
On Fri, Nov 14, 2008 at 03:43:08PM +0100, Farkas Levente wrote:
currently koji able to build any of the mingw32-x packages? i means it's know that he has to use *-bootstrap packages?
We will negotiate this with Fedora rel-eng when the time comes to build the packages.
Rich.
On Tue, Nov 11, 2008 at 12:45:40AM +0100, Farkas Levente wrote:
i686-pc-mingw32-gcc -c -I/usr/i686-pc-mingw32/sys-root/mingw/include -Wall -fomit-frame-pointer -I. -I./.. -I./../include -nostdinc -iwithprefixbefore include -I ./../../w32api/include usleep.c -o usleep.o usleep.c:32: error: redefinition of typedef 'useconds_t' /usr/i686-pc-mingw32/sys-root/mingw/include/sys/types.h:119: error: previous declaration of 'useconds_t' was here
Now I see this problem ... on Fedora 10 when building against runtime 3.15.
I've removed my patch and it builds.
Rich.
Richard W.M. Jones wrote:
On Tue, Nov 11, 2008 at 12:45:40AM +0100, Farkas Levente wrote:
i686-pc-mingw32-gcc -c -I/usr/i686-pc-mingw32/sys-root/mingw/include -Wall -fomit-frame-pointer -I. -I./.. -I./../include -nostdinc -iwithprefixbefore include -I ./../../w32api/include usleep.c -o usleep.o usleep.c:32: error: redefinition of typedef 'useconds_t' /usr/i686-pc-mingw32/sys-root/mingw/include/sys/types.h:119: error: previous declaration of 'useconds_t' was here
Now I see this problem ... on Fedora 10 when building against runtime 3.15.
I've removed my patch and it builds.
did you made a new/updated bootstrap packages? if not it's on my todo list...
On Wed, Nov 19, 2008 at 08:38:05PM +0100, Farkas Levente wrote:
Richard W.M. Jones wrote:
On Tue, Nov 11, 2008 at 12:45:40AM +0100, Farkas Levente wrote:
i686-pc-mingw32-gcc -c -I/usr/i686-pc-mingw32/sys-root/mingw/include -Wall -fomit-frame-pointer -I. -I./.. -I./../include -nostdinc -iwithprefixbefore include -I ./../../w32api/include usleep.c -o usleep.o usleep.c:32: error: redefinition of typedef 'useconds_t' /usr/i686-pc-mingw32/sys-root/mingw/include/sys/types.h:119: error: previous declaration of 'useconds_t' was here
Now I see this problem ... on Fedora 10 when building against runtime 3.15.
I've removed my patch and it builds.
did you made a new/updated bootstrap packages? if not it's on my todo list...
I didn't, but it doesn't really matter now because we can just use the mingw32-w32api/mingw32-runtime binaries, and if rel-eng won't do those then we can update the bootstrap packages down the road when we are adding them to Fedora.
Rich.
Richard W.M. Jones wrote:
On Wed, Nov 19, 2008 at 08:38:05PM +0100, Farkas Levente wrote:
Richard W.M. Jones wrote:
On Tue, Nov 11, 2008 at 12:45:40AM +0100, Farkas Levente wrote:
i686-pc-mingw32-gcc -c -I/usr/i686-pc-mingw32/sys-root/mingw/include -Wall -fomit-frame-pointer -I. -I./.. -I./../include -nostdinc -iwithprefixbefore include -I ./../../w32api/include usleep.c -o usleep.o usleep.c:32: error: redefinition of typedef 'useconds_t' /usr/i686-pc-mingw32/sys-root/mingw/include/sys/types.h:119: error: previous declaration of 'useconds_t' was here
Now I see this problem ... on Fedora 10 when building against runtime 3.15.
I've removed my patch and it builds.
did you made a new/updated bootstrap packages? if not it's on my todo list...
I didn't, but it doesn't really matter now because we can just use the mingw32-w32api/mingw32-runtime binaries, and if rel-eng won't do those then we can update the bootstrap packages down the road when we are adding them to Fedora.
i'm not rally understand you. without the bootstrap packages how can you build mingw32-runtime in mock?
On Wed, Nov 19, 2008 at 08:59:18PM +0100, Farkas Levente wrote:
i'm not rally understand you. without the bootstrap packages how can you build mingw32-runtime in mock?
By copying the built mingw32-w32api / mingw32-runtime binary RPMs from another machine. They are 'noarch' so that even works for bootstrapping ppc and other non-x86 platforms.
But if we need to build the *-bootstrap packages for rel-eng we can do that later, if they want it.
Rich.