| Message ID | 1491568417-5355-1-git-send-email-heiko.thiery@gmail.com |
|---|---|
| State | Changes Requested |
| Headers | show |
2017-04-07 14:33 GMT+02:00 Heiko Thiery <heiko.thiery@gmail.com>: > A bug in automake makes it impossible to build in an output directory > that contains an '@' character. But this is needed if you want to build > with jenkins' pipeline plugin. > > This patch was proposed by Michael Brantley but is currently not applied > in the upstream automake repository. > > Refer to: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25299 > > Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com> > --- > ...character-in-installation-prefix-breaks-automake.patch | 15 > +++++++++++++++ > 1 file changed, 15 insertions(+) > create mode 100644 package/automake/0003-at-character-in-installation- > prefix-breaks-automake.patch > > diff --git a/package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch > b/package/automake/0003-at-character-in-installation- > prefix-breaks-automake.patch > new file mode 100644 > index 0000000..3a5a571 > --- /dev/null > +++ b/package/automake/0003-at-character-in-installation- > prefix-breaks-automake.patch > @@ -0,0 +1,15 @@ > +See bug report #25299 (https://debbugs.gnu.org/cgi/ > bugreport.cgi?bug=25299) > + > +Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com> > + > +--- a/bin/aclocal.in > ++++ b/bin/aclocal.in > +@@ -69,7 +69,7 @@ > + # ACLOCAL_PATH environment variable, and reset with the '--system-acdir' > + # option. > + my @user_includes = (); > +-my @automake_includes = ("@datadir@/aclocal-$APIVERSION"); > ++my @automake_includes = ('@datadir@/aclocal-' . $APIVERSION); > + my @system_includes = ('@datadir@/aclocal'); > + > + # Whether we should copy M4 file in $user_includes[0]. > -- > 2.1.4 > > Can someone give me some feedback if this patch is ok and will be accepted? BR, Heiko
On 10-04-17 07:48, Heiko Thiery wrote: > 2017-04-07 14:33 GMT+02:00 Heiko Thiery <heiko.thiery@gmail.com > <mailto:heiko.thiery@gmail.com>>: > > A bug in automake makes it impossible to build in an output directory > that contains an '@' character. But this is needed if you want to build > with jenkins' pipeline plugin. > > This patch was proposed by Michael Brantley but is currently not applied > in the upstream automake repository. > > Refer to: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25299 > <https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25299> > > Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com > <mailto:heiko.thiery@gmail.com>> > --- > ...character-in-installation-prefix-breaks-automake.patch | 15 +++++++++++++++ > 1 file changed, 15 insertions(+) > create mode 100644 > package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch > > diff --git > a/package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch > b/package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch > new file mode 100644 > index 0000000..3a5a571 > --- /dev/null > +++ > b/package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch > @@ -0,0 +1,15 @@ > +See bug report #25299 (https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25299 > <https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25299>) > + > +Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com > <mailto:heiko.thiery@gmail.com>> > + > +--- a/bin/aclocal.in <http://aclocal.in> > ++++ b/bin/aclocal.in <http://aclocal.in> > +@@ -69,7 +69,7 @@ > + # ACLOCAL_PATH environment variable, and reset with the '--system-acdir' > + # option. > + my @user_includes = (); > +-my @automake_includes = ("@datadir@/aclocal-$APIVERSION"); > ++my @automake_includes = ('@datadir@/aclocal-' . $APIVERSION); > + my @system_includes = ('@datadir@/aclocal'); > + > + # Whether we should copy M4 file in $user_includes[0]. > -- > 2.1.4 > > > Can someone give me some feedback if this patch is ok and will be accepted? You're a bit impatient, we have many patches pending review (cfr. http://patchwork.ozlabs.org/project/buildroot/list/ ). That said, chances are pretty low that anybody would look at this patch. The problem is that it's a kind of feature patch (it doesn't solve a cross-compilation issue or regression), and it's not easy to understand if it does the right thing. So we would typically want upstream feedback on it. Unfortunately, there is no activity on the bug since 4 months (except for your +1). One thing that would help is if your commit message would clarify that the bug only shows itself on packages that are autoreconf'ed, and a minimal defconfig that allows someone to reproduce. Something like: BR2_TOOLCHAIN_EXTERNAL=y BR2_PACKAGE_LZO=y make O=@output olddefconfig world And it would also help if you could give a list of autoreconf'ed packages that you have tested with this patch. Regards, Arnout
Hello, On Mon, 10 Apr 2017 16:10:06 +0200, Arnout Vandecappelle wrote: > And it would also help if you could give a list of autoreconf'ed packages that > you have tested with this patch. I thought I had given feedback on this patch, but apparently no. One think that isn't clear to me is what packages are affected: * Are only packages autoreconf'ed by Buildroot affected by the @ issue ? * Or are all autotools packages affected ? If that's the case, then this patch is not sufficient: we would have to autoreconf all autotools packages, but is not really doable. Best regards, Thomas
On 14-04-17 22:43, Thomas Petazzoni wrote: > Hello, > > On Mon, 10 Apr 2017 16:10:06 +0200, Arnout Vandecappelle wrote: > >> And it would also help if you could give a list of autoreconf'ed packages that >> you have tested with this patch. > > I thought I had given feedback on this patch, but apparently no. One > think that isn't clear to me is what packages are affected: > > * Are only packages autoreconf'ed by Buildroot affected by the @ > issue ? > > * Or are all autotools packages affected ? If that's the case, then > this patch is not sufficient: we would have to autoreconf all > autotools packages, but is not really doable. I've tried it, and apparently only autoreconf'ed packages are the problem. That's why I wrote that the commit message should clarify exactly this. Regards, Arnout
diff --git a/package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch b/package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch new file mode 100644 index 0000000..3a5a571 --- /dev/null +++ b/package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch @@ -0,0 +1,15 @@ +See bug report #25299 (https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25299) + +Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com> + +--- a/bin/aclocal.in ++++ b/bin/aclocal.in +@@ -69,7 +69,7 @@ + # ACLOCAL_PATH environment variable, and reset with the '--system-acdir' + # option. + my @user_includes = (); +-my @automake_includes = ("@datadir@/aclocal-$APIVERSION"); ++my @automake_includes = ('@datadir@/aclocal-' . $APIVERSION); + my @system_includes = ('@datadir@/aclocal'); + + # Whether we should copy M4 file in $user_includes[0].
A bug in automake makes it impossible to build in an output directory that contains an '@' character. But this is needed if you want to build with jenkins' pipeline plugin. This patch was proposed by Michael Brantley but is currently not applied in the upstream automake repository. Refer to: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25299 Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com> --- ...character-in-installation-prefix-breaks-automake.patch | 15 +++++++++++++++ 1 file changed, 15 insertions(+) create mode 100644 package/automake/0003-at-character-in-installation-prefix-breaks-automake.patch