| Submitter | Steve Sakoman |
|---|---|
| Date | Feb. 21, 2012, 7:07 p.m. |
| Message ID | <1329851229-31625-1-git-send-email-steve@sakoman.com> |
| Download | mbox | patch |
| Permalink | /patch/21497/ |
| State | New |
| Headers | show |
Comments
On Tue, 2012-02-21 at 11:07 -0800, Steve Sakoman wrote: > Otherwise build fails with following error: > > | In file included from libxml.h:36:0, > | from xmlmemory.c:8: > | ./include/libxml/xmlversion.h:396:22: fatal error: ansidecl.h: No such file or directory > > Signed-off-by: Steve Sakoman <steve@sakoman.com> > --- > meta/recipes-core/libxml/libxml2.inc | 2 +- > 1 files changed, 1 insertions(+), 1 deletions(-) > > diff --git a/meta/recipes-core/libxml/libxml2.inc b/meta/recipes-core/libxml/libxml2.inc > index 0979acd..6230f60 100644 > --- a/meta/recipes-core/libxml/libxml2.inc > +++ b/meta/recipes-core/libxml/libxml2.inc > @@ -10,7 +10,7 @@ LIC_FILES_CHKSUM = "file://Copyright;md5=bb90c48926316d9af6e2d70ca7013ade \ > file://trio.c;beginline=5;endline=14;md5=6c025753c86d958722ec76e94cae932e" > > DEPENDS_virtclass-native = "python-native" > -DEPENDS =+ "zlib" > +DEPENDS =+ "zlib binutils" > > SRC_URI = "ftp://xmlsoft.org/libxml2/libxml2-${PV}.tar.gz \ > file://libxml-64bit.patch" This sounds very wrong. It would have implications for libxml2-native I'd imagine. What is it doing that needs binutils? Its going to add considerable time to the build too. Cheers, Richard
On Tue, Feb 21, 2012 at 1:52 PM, Richard Purdie <richard.purdie@linuxfoundation.org> wrote: > On Tue, 2012-02-21 at 11:07 -0800, Steve Sakoman wrote: >> Otherwise build fails with following error: >> >> | In file included from libxml.h:36:0, >> | from xmlmemory.c:8: >> | ./include/libxml/xmlversion.h:396:22: fatal error: ansidecl.h: No such file or directory >> >> Signed-off-by: Steve Sakoman <steve@sakoman.com> >> --- >> meta/recipes-core/libxml/libxml2.inc | 2 +- >> 1 files changed, 1 insertions(+), 1 deletions(-) >> >> diff --git a/meta/recipes-core/libxml/libxml2.inc b/meta/recipes-core/libxml/libxml2.inc >> index 0979acd..6230f60 100644 >> --- a/meta/recipes-core/libxml/libxml2.inc >> +++ b/meta/recipes-core/libxml/libxml2.inc >> @@ -10,7 +10,7 @@ LIC_FILES_CHKSUM = "file://Copyright;md5=bb90c48926316d9af6e2d70ca7013ade \ >> file://trio.c;beginline=5;endline=14;md5=6c025753c86d958722ec76e94cae932e" >> >> DEPENDS_virtclass-native = "python-native" >> -DEPENDS =+ "zlib" >> +DEPENDS =+ "zlib binutils" >> >> SRC_URI = "ftp://xmlsoft.org/libxml2/libxml2-${PV}.tar.gz \ >> file://libxml-64bit.patch" > > This sounds very wrong. It would have implications for libxml2-native > I'd imagine. What is it doing that needs binutils? For some reason the libxml2 configure phase decides that ansidecl.h from binutils is present, but then at compile time it finds that it actually isn't there! I saw the same issue on both my desktop and autobuilder setups. I'll try a clean build to see if the problem persists. The libz change from last week seems to have put my builds in a bad state, and the PR bumps intended to fix things really don't seem to be fixing things, at least on my build machines. I suspect it might be faster to do a clean build than to track down where the corruption is. Steve
On Tue, Feb 21, 2012 at 02:26:49PM -0800, Steve Sakoman wrote: > On Tue, Feb 21, 2012 at 1:52 PM, Richard Purdie > <richard.purdie@linuxfoundation.org> wrote: > > On Tue, 2012-02-21 at 11:07 -0800, Steve Sakoman wrote: > >> Otherwise build fails with following error: > >> > >> | In file included from libxml.h:36:0, > >> | from xmlmemory.c:8: > >> | ./include/libxml/xmlversion.h:396:22: fatal error: ansidecl.h: No such file or directory > >> > >> Signed-off-by: Steve Sakoman <steve@sakoman.com> > >> --- > >> meta/recipes-core/libxml/libxml2.inc | 2 +- > >> 1 files changed, 1 insertions(+), 1 deletions(-) > >> > >> diff --git a/meta/recipes-core/libxml/libxml2.inc b/meta/recipes-core/libxml/libxml2.inc > >> index 0979acd..6230f60 100644 > >> --- a/meta/recipes-core/libxml/libxml2.inc > >> +++ b/meta/recipes-core/libxml/libxml2.inc > >> @@ -10,7 +10,7 @@ LIC_FILES_CHKSUM = "file://Copyright;md5=bb90c48926316d9af6e2d70ca7013ade \ > >> file://trio.c;beginline=5;endline=14;md5=6c025753c86d958722ec76e94cae932e" > >> > >> DEPENDS_virtclass-native = "python-native" > >> -DEPENDS =+ "zlib" > >> +DEPENDS =+ "zlib binutils" > >> > >> SRC_URI = "ftp://xmlsoft.org/libxml2/libxml2-${PV}.tar.gz \ > >> file://libxml-64bit.patch" > > > > This sounds very wrong. It would have implications for libxml2-native > > I'd imagine. What is it doing that needs binutils? > > For some reason the libxml2 configure phase decides that ansidecl.h > from binutils is present, but then at compile time it finds that it > actually isn't there! I saw the same issue on both my desktop and > autobuilder setups. > > I'll try a clean build to see if the problem persists. The libz > change from last week seems to have put my builds in a bad state, and > the PR bumps intended to fix things really don't seem to be fixing > things, at least on my build machines. I suspect it might be faster > to do a clean build than to track down where the corruption is. The list of PR bumped recipes probably wasn't complete for you. It was for me, but later I've noticed that ie subversion-1.6 wasn't bumped, because I'm using only 1.7. So it was complete for my default build (when those PR bumps were applied to all included layers at once, which is not the case for most people) and if you build different recipes or different versions you have to add your own PR bumps in those recipes (see khem's script). Clean build could be faster but doesn't fix installed -dev packages on target.. Cheers,
On Tue, Feb 21, 2012 at 1:52 PM, Richard Purdie <richard.purdie@linuxfoundation.org> wrote: > On Tue, 2012-02-21 at 11:07 -0800, Steve Sakoman wrote: >> Otherwise build fails with following error: >> >> | In file included from libxml.h:36:0, >> | from xmlmemory.c:8: >> | ./include/libxml/xmlversion.h:396:22: fatal error: ansidecl.h: No such file or directory >> >> Signed-off-by: Steve Sakoman <steve@sakoman.com> >> --- >> meta/recipes-core/libxml/libxml2.inc | 2 +- >> 1 files changed, 1 insertions(+), 1 deletions(-) >> >> diff --git a/meta/recipes-core/libxml/libxml2.inc b/meta/recipes-core/libxml/libxml2.inc >> index 0979acd..6230f60 100644 >> --- a/meta/recipes-core/libxml/libxml2.inc >> +++ b/meta/recipes-core/libxml/libxml2.inc >> @@ -10,7 +10,7 @@ LIC_FILES_CHKSUM = "file://Copyright;md5=bb90c48926316d9af6e2d70ca7013ade \ >> file://trio.c;beginline=5;endline=14;md5=6c025753c86d958722ec76e94cae932e" >> >> DEPENDS_virtclass-native = "python-native" >> -DEPENDS =+ "zlib" >> +DEPENDS =+ "zlib binutils" >> >> SRC_URI = "ftp://xmlsoft.org/libxml2/libxml2-${PV}.tar.gz \ >> file://libxml-64bit.patch" > > This sounds very wrong. It would have implications for libxml2-native > I'd imagine. What is it doing that needs binutils? configure detects the header and then uses it. So as long as you provide binutils or not provide consistently it will build in a consistent way whats happening here probably is that its configured with binutils-dev being staged but is building without then being in staging. > > Its going to add considerable time to the build too. > > Cheers, > > Richard > > > _______________________________________________ > Openembedded-core mailing list > Openembedded-core@lists.openembedded.org > http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
On Tue, 2012-02-21 at 23:48 +0100, Martin Jansa wrote: > On Tue, Feb 21, 2012 at 02:26:49PM -0800, Steve Sakoman wrote: > > I'll try a clean build to see if the problem persists. The libz > > change from last week seems to have put my builds in a bad state, and > > the PR bumps intended to fix things really don't seem to be fixing > > things, at least on my build machines. I suspect it might be faster > > to do a clean build than to track down where the corruption is. > > The list of PR bumped recipes probably wasn't complete for you. It was > for me, but later I've noticed that ie subversion-1.6 wasn't bumped, > because I'm using only 1.7. > > So it was complete for my default build (when those PR bumps were > applied to all included layers at once, which is not the case for most > people) and if you build different recipes or different versions you > have to add your own PR bumps in those recipes (see khem's script). > > Clean build could be faster but doesn't fix installed -dev packages on > target.. I still don't entirely understand why we don't just put libz.la back the way it was. I can understand the attraction of eliminating .la files in a general sense, but it doesn't seem as though the removal of this particular one is buying us much. What is the reason for expending so much effort on getting rid of it? p.
On Tue, Feb 21, 2012 at 2:26 PM, Steve Sakoman <steve@sakoman.com> wrote: > On Tue, Feb 21, 2012 at 1:52 PM, Richard Purdie > <richard.purdie@linuxfoundation.org> wrote: >> On Tue, 2012-02-21 at 11:07 -0800, Steve Sakoman wrote: >>> Otherwise build fails with following error: >>> >>> | In file included from libxml.h:36:0, >>> | from xmlmemory.c:8: >>> | ./include/libxml/xmlversion.h:396:22: fatal error: ansidecl.h: No such file or directory >>> >>> Signed-off-by: Steve Sakoman <steve@sakoman.com> >>> --- >>> meta/recipes-core/libxml/libxml2.inc | 2 +- >>> 1 files changed, 1 insertions(+), 1 deletions(-) >>> >>> diff --git a/meta/recipes-core/libxml/libxml2.inc b/meta/recipes-core/libxml/libxml2.inc >>> index 0979acd..6230f60 100644 >>> --- a/meta/recipes-core/libxml/libxml2.inc >>> +++ b/meta/recipes-core/libxml/libxml2.inc >>> @@ -10,7 +10,7 @@ LIC_FILES_CHKSUM = "file://Copyright;md5=bb90c48926316d9af6e2d70ca7013ade \ >>> file://trio.c;beginline=5;endline=14;md5=6c025753c86d958722ec76e94cae932e" >>> >>> DEPENDS_virtclass-native = "python-native" >>> -DEPENDS =+ "zlib" >>> +DEPENDS =+ "zlib binutils" >>> >>> SRC_URI = "ftp://xmlsoft.org/libxml2/libxml2-${PV}.tar.gz \ >>> file://libxml-64bit.patch" >> >> This sounds very wrong. It would have implications for libxml2-native >> I'd imagine. What is it doing that needs binutils? > > For some reason the libxml2 configure phase decides that ansidecl.h > from binutils is present, but then at compile time it finds that it > actually isn't there! I saw the same issue on both my desktop and > autobuilder setups. > > I'll try a clean build to see if the problem persists. The libz > change from last week seems to have put my builds in a bad state, and > the PR bumps intended to fix things really don't seem to be fixing > things, at least on my build machines. I suspect it might be faster > to do a clean build than to track down where the corruption is. FWIW, a clean build resolved the binutils dependency, so the above patch is not required. I suspect that the current set of PR bumps for the zlib issue still aren't sufficient for many users, but I'm hoping the switch to OEBasicHash will minimize further pain. Steve
Patch
diff --git a/meta/recipes-core/libxml/libxml2.inc b/meta/recipes-core/libxml/libxml2.inc index 0979acd..6230f60 100644 --- a/meta/recipes-core/libxml/libxml2.inc +++ b/meta/recipes-core/libxml/libxml2.inc @@ -10,7 +10,7 @@ LIC_FILES_CHKSUM = "file://Copyright;md5=bb90c48926316d9af6e2d70ca7013ade \ file://trio.c;beginline=5;endline=14;md5=6c025753c86d958722ec76e94cae932e" DEPENDS_virtclass-native = "python-native" -DEPENDS =+ "zlib" +DEPENDS =+ "zlib binutils" SRC_URI = "ftp://xmlsoft.org/libxml2/libxml2-${PV}.tar.gz \ file://libxml-64bit.patch"
Otherwise build fails with following error: | In file included from libxml.h:36:0, | from xmlmemory.c:8: | ./include/libxml/xmlversion.h:396:22: fatal error: ansidecl.h: No such file or directory Signed-off-by: Steve Sakoman <steve@sakoman.com> --- meta/recipes-core/libxml/libxml2.inc | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)