| Submitter | Ross Burton |
|---|---|
| Date | Sept. 10, 2012, 4:02 p.m. |
| Message ID | <1347292942-7241-1-git-send-email-ross.burton@intel.com> |
| Download | mbox | patch |
| Permalink | /patch/36215/ |
| State | Accepted |
| Commit | af5bdc8ca413d6cabeb4f4b4c5836912a17f28be |
| Headers | show |
Comments
On Mon, 2012-09-10 at 17:02 +0100, Ross Burton wrote:
> +# fixed in Make CVS, so 3.83 won't have this problem.
I know this will be painful, but did you consider looking at the make
version and determine whether or not this workaround is necessary?
On Mon, 2012-09-10 at 17:14 -0400, Colin Walters wrote: > On Mon, 2012-09-10 at 17:02 +0100, Ross Burton wrote: > > > +# fixed in Make CVS, so 3.83 won't have this problem. > > I know this will be painful, but did you consider looking at the make > version and determine whether or not this workaround is necessary? Or actually, why not just backport the make patch into core now?
On Mon, 2012-09-10 at 17:23 -0400, Colin Walters wrote: > On Mon, 2012-09-10 at 17:14 -0400, Colin Walters wrote: > > On Mon, 2012-09-10 at 17:02 +0100, Ross Burton wrote: > > > > > +# fixed in Make CVS, so 3.83 won't have this problem. > > > > I know this will be painful, but did you consider looking at the make > > version and determine whether or not this workaround is necessary? > > Or actually, why not just backport the make patch into core now? We've gone around in circles on this several times. The version check complicates the code more than I'd like. Adding a make-native causes its own sets of pain and hoops we'd need to jump through. So I'm probably at the point I'll take this solution since its the "least worst" :/. Cheers, Richard
On 10 September 2012 22:23, Colin Walters <walters@verbum.org> wrote: > On Mon, 2012-09-10 at 17:14 -0400, Colin Walters wrote: >> On Mon, 2012-09-10 at 17:02 +0100, Ross Burton wrote: >> >> > +# fixed in Make CVS, so 3.83 won't have this problem. >> >> I know this will be painful, but did you consider looking at the make >> version and determine whether or not this workaround is necessary? > > Or actually, why not just backport the make patch into core now? I had a branch that added make-replacement-native 3.81 as a dependency of WebKit, but that's far too fragile. An alternative would be to add make-native as part of the bootstrap. At this late stage in the release that's rather invasive. Ross
On 09/10/2012 09:02 AM, Ross Burton wrote: > GNU make 3.82 has a bug where it drops required dependencies. > https://bugs.webkit.org/show_bug.cgi?id=79498 is the WebKitGTK+ > bug, and http://savannah.gnu.org/bugs/?30653 is the GNU Make bug. > > Work around this by running make again if it fails just in case the failure is > due to the bug. > > Based on a patch by Andreas Müller <schnitzeltony@googlemail.com>. > > [ YOCTO #2816 ] > > Signed-off-by: Ross Burton <ross.burton@intel.com> > --- > meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb | 22 ++++++++++++++++++++++ > 1 file changed, 22 insertions(+) > Merged into OE-Core Thanks Sau! > diff --git a/meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb b/meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb > index cf4129d..8beb6eb 100644 > --- a/meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb > +++ b/meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb > @@ -7,6 +7,8 @@ LIC_FILES_CHKSUM = "file://Source/WebCore/rendering/RenderApplet.h;endline=22;md > file://Source/WebKit/gtk/webkit/webkit.h;endline=21;md5=b4fbe9f4a944f1d071dba1d2c76b3351 \ > file://Source/JavaScriptCore/parser/Parser.h;endline=23;md5=b57c8a2952a8d0e655988fa0ecb2bf7f" > > +PR = "r1" > + > # Choice of language backends - icu has issues on Big Endian machines so use pango > ICU_LIB = "icu" > ICU_LIB_powerpc = "pango" > @@ -77,6 +79,26 @@ do_configure_append() { > done > } > > +# A dirty hack for GNU make 3.82 bug which means it drops required > +# dependencies. https://bugs.webkit.org/show_bug.cgi?id=79498 is the WebKitGTK+ > +# bug, and http://savannah.gnu.org/bugs/?30653 is the GNU Make bug. This is > +# fixed in Make CVS, so 3.83 won't have this problem. > +do_compile() { > + if [ x"$MAKE" = x ]; then MAKE=make; fi > + bbnote ${MAKE} ${EXTRA_OEMAKE} "$@" > + for error_count in 1 2 3; do > + bbnote "Attempt $error_count of 3" > + exit_code=0 > + ${MAKE} ${EXTRA_OEMAKE} "$@" || exit_code=1 > + if [ $exit_code = 0 ]; then > + break > + fi > + done > + if [ ! $exit_code = 0 ]; then > + die "oe_runmake failed" > + fi > +} > + > do_install_append() { > rmdir ${D}${libexecdir} > } >
Patch
diff --git a/meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb b/meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb index cf4129d..8beb6eb 100644 --- a/meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb +++ b/meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb @@ -7,6 +7,8 @@ LIC_FILES_CHKSUM = "file://Source/WebCore/rendering/RenderApplet.h;endline=22;md file://Source/WebKit/gtk/webkit/webkit.h;endline=21;md5=b4fbe9f4a944f1d071dba1d2c76b3351 \ file://Source/JavaScriptCore/parser/Parser.h;endline=23;md5=b57c8a2952a8d0e655988fa0ecb2bf7f" +PR = "r1" + # Choice of language backends - icu has issues on Big Endian machines so use pango ICU_LIB = "icu" ICU_LIB_powerpc = "pango" @@ -77,6 +79,26 @@ do_configure_append() { done } +# A dirty hack for GNU make 3.82 bug which means it drops required +# dependencies. https://bugs.webkit.org/show_bug.cgi?id=79498 is the WebKitGTK+ +# bug, and http://savannah.gnu.org/bugs/?30653 is the GNU Make bug. This is +# fixed in Make CVS, so 3.83 won't have this problem. +do_compile() { + if [ x"$MAKE" = x ]; then MAKE=make; fi + bbnote ${MAKE} ${EXTRA_OEMAKE} "$@" + for error_count in 1 2 3; do + bbnote "Attempt $error_count of 3" + exit_code=0 + ${MAKE} ${EXTRA_OEMAKE} "$@" || exit_code=1 + if [ $exit_code = 0 ]; then + break + fi + done + if [ ! $exit_code = 0 ]; then + die "oe_runmake failed" + fi +} + do_install_append() { rmdir ${D}${libexecdir} }
GNU make 3.82 has a bug where it drops required dependencies. https://bugs.webkit.org/show_bug.cgi?id=79498 is the WebKitGTK+ bug, and http://savannah.gnu.org/bugs/?30653 is the GNU Make bug. Work around this by running make again if it fails just in case the failure is due to the bug. Based on a patch by Andreas Müller <schnitzeltony@googlemail.com>. [ YOCTO #2816 ] Signed-off-by: Ross Burton <ross.burton@intel.com> --- meta/recipes-sato/webkit/webkit-gtk_1.8.2.bb | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+)