Skip to contentSkip to conversation

fix parallel build failures in xfsprogs-3.0.0

PATCH In review Mike Frysinger 6 participants 43 messages Feb 24, 2009 – Mar 27, 2009 Kbuild

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fix parallel build failures in xfsprogs-3.0.0

i noticed that xfsprogs-3.0.0 sometimes fails to build in parallel with an error about fsr not being able to find libhandle. looking at the top level Makefile shows the obvious missing depend (fsr: libhandle). playing around a bit found a few more not so obvious missing depends on the include subdir, so i fixed that too.
-mike

http://bugs.gentoo.org/260005

Makefile +4 -3
@@ -15,11 +15,11 @@
 LDIRT = config.log .dep config.status config.cache confdefs.h conftest* \
 	Logs/* built .census install.* install-dev.* *.gz
 
-LIB_SUBDIRS = include libxfs libxlog libxcmd libhandle libdisk
+LIB_SUBDIRS = libxfs libxlog libxcmd libhandle libdisk
 TOOL_SUBDIRS = copy db estimate fsck fsr growfs io logprint mkfs quota \
 		mdrestore repair rtcp m4 man doc po debian build
 
-SUBDIRS = $(LIB_SUBDIRS) $(TOOL_SUBDIRS)
+SUBDIRS = include $(LIB_SUBDIRS) $(TOOL_SUBDIRS)
 
 default: include/builddefs include/platform_defs.h
 ifeq ($(HAVE_BUILDDEFS), no)
@@ -29,9 +29,10 @@
 endif
 
 # tool/lib dependencies
-libxcmd: include
+$(LIB_SUBDIRS) $(TOOL_SUBDIRS): include
 copy mdrestore: libxfs
 db logprint: libxfs libxlog
+fsr: libhandle
 growfs: libxfs libxcmd
 io: libxcmd libhandle
 mkfs: libxfs libdisk

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#1 lore raw
Quoted 6 lines · Mike Frysinger

i noticed that xfsprogs-3.0.0 sometimes fails to build in parallel with an error about fsr not being able to find libhandle. looking at the top level Makefile shows the obvious missing depend (fsr: libhandle). playing around a bit found a few more not so obvious missing depends on the include subdir, so i fixed that too.
-mike

Seems fine. Looks like that dep got missed when fsr moved.

the include dependency might be a big hammer, but *shrug* (it takes care of the "estimate" dependency on include too)

I'll check this in, thanks.

-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#2 lore raw
In reply to Mike Frysinger
Quoted 6 lines · Mike Frysinger

i noticed that xfsprogs-3.0.0 sometimes fails to build in parallel with an error about fsr not being able to find libhandle. looking at the top level Makefile shows the obvious missing depend (fsr: libhandle). playing around a bit found a few more not so obvious missing depends on the include subdir, so i fixed that too.
-mike

Oh, can you resend with a formal Signed-off-by: line? Being a little nitpicky here but we follow the same process as kernel patches.

Thanks,
-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#3 lore raw
Quoted 9 lines · On Tuesday 24 February 2009 00:22:40 Eric Sandeen wrote:
Quoted 5 lines · Mike Frysinger

i noticed that xfsprogs-3.0.0 sometimes fails to build in parallel with an error about fsr not being able to find libhandle. looking at the top level Makefile shows the obvious missing depend (fsr: libhandle). playing around a bit found a few more not so obvious missing depends on the include subdir, so i fixed that too.

Oh, can you resend with a formal Signed-off-by: line? Being a little nitpicky here but we follow the same process as kernel patches.

where is the git tree now ? i tried git://oss.sgi.com/xfs-cmds but that seems to be dead now.
-mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#4 lore raw
Quoted 8 lines · Mike Frysinger
Quoted 7 lines · On Tuesday 24 February 2009 00:22:40 Eric Sandeen wrote:
Quoted 5 lines · Mike Frysinger

i noticed that xfsprogs-3.0.0 sometimes fails to build in parallel with an error about fsr not being able to find libhandle. looking at the top level Makefile shows the obvious missing depend (fsr: libhandle). playing around a bit found a few more not so obvious missing depends on the include subdir, so i fixed that too.

Quoted 5 lines
Quoted 2 lines

Oh, can you resend with a formal Signed-off-by: line? Being a little nitpicky here but we follow the same process as kernel patches.

where is the git tree now ? i tried git://oss.sgi.com/xfs-cmds but that seems to be dead now.

http://git.kernel.org/?p=fs/xfs/xfsprogs-dev.git;a=tree for the bleeding edge development tree.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#5 lore raw
Quoted 9 lines · On Tuesday 24 February 2009 08:15:23 Christoph Hellwig wrote:
Quoted 8 lines · Mike Frysinger
Quoted 7 lines · On Tuesday 24 February 2009 00:22:40 Eric Sandeen wrote:
Quoted 5 lines · Mike Frysinger

i noticed that xfsprogs-3.0.0 sometimes fails to build in parallel with an error about fsr not being able to find libhandle. looking at the top level Makefile shows the obvious missing depend (fsr:
libhandle). playing around a bit found a few more not so obvious missing depends on the include subdir, so i fixed that too.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

Oh, can you resend with a formal Signed-off-by: line? Being a little nitpicky here but we follow the same process as kernel patches.

where is the git tree now ? i tried git://oss.sgi.com/xfs-cmds but that seems to be dead now.

http://git.kernel.org/?p=fs/xfs/xfsprogs-dev.git;a=tree for the bleeding edge development tree.

ok, so i'll base all my patches on the git trees that are available on kernel.org.

the xfs.org wiki talks about git trees on sgi.com, but some of them are old and/or moved (like the xfs-cmds one) ...
-mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#6 lore raw
Quoted 1 line · Mike Frysinger

the xfs.org wiki talks about git trees on sgi.com,

Yes, they're still there.

Quoted 2 lines

but some of them are old
and/or moved (like the xfs-cmds one) ...

xfs-cmds was split into separate repos for xfsprogs, xfsdump, and dmapi. And that info is on xfs.org. Here is the excerpt from the following page:

http://www.xfs.org/index.php/Getting_the_latest_source_code

XFS user space tools

* xfsprogs

$ git clone git://oss.sgi.com/xfs/cmds/xfsprogs

XFS dump

* xfsdump

$ git clone git://oss.sgi.com/xfs/cmds/xfsdump

XFS tests

* xfstests

$ git clone git://oss.sgi.com/xfs/cmds/xfstests

DMAPI user space tools

* dmapi

$ git clone git://oss.sgi.com/xfs/cmds/dmapi

At times though the repos on kernel.org could be ahead of sgi ones.

Felix

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#7 lore raw
Quoted 10 lines · On Tuesday 24 February 2009 10:14:13 Felix Blyakher wrote:
Quoted 1 line · Mike Frysinger

the xfs.org wiki talks about git trees on sgi.com,

Yes, they're still there.

Quoted 2 lines

but some of them are old
and/or moved (like the xfs-cmds one) ...

xfs-cmds was split into separate repos for xfsprogs, xfsdump, and dmapi. And that info is on xfs.org.

but acl/attr have not. distros care about those packages as well.

Quoted 1 line

Here is the excerpt from the following page:

yes, i was reading the page. i was also pointing out the broken link to the xfs-cmds repo (which is after everything you quoted) and that that repo seems to be dead now. which means there is no up-to-date place for acl/attr on that page.

Quoted 1 line

At times though the repos on kernel.org could be ahead of sgi ones.

well if people are going to try to help in development, the kernel.org trees would be good to mention on that page as well with a note as to which tree is preferred (sgi.com or kernel.org).
-mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#8 lore raw
Quoted 6 lines · Mike Frysinger
Quoted 5 lines · On Tuesday 24 February 2009 10:14:13 Felix Blyakher wrote:
Quoted 1 line · Mike Frysinger

the xfs.org wiki talks about git trees on sgi.com,

Yes, they're still there.

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

but some of them are old
and/or moved (like the xfs-cmds one) ...

xfs-cmds was split into separate repos for xfsprogs, xfsdump, and dmapi. And that info is on xfs.org.

but acl/attr have not. distros care about those packages as well.

Andreas who has done most of the maintaince already is taking these over formally, so they won't be released with the xfs tools anymore. I've put him on Cc so he can say when he expects to clear the backlog.

Quoted 6 lines
Quoted 1 line

Here is the excerpt from the following page:

yes, i was reading the page. i was also pointing out the broken link to the xfs-cmds repo (which is after everything you quoted) and that that repo seems to be dead now. which means there is no up-to-date place for acl/attr on that page.

Right now the most uptodate acl packages are the on on kernel.org, but they will move (probably still on kernel.org, but not under fs/xfs)

Quoted 5 lines
Quoted 1 line

At times though the repos on kernel.org could be ahead of sgi ones.

well if people are going to try to help in development, the kernel.org trees would be good to mention on that page as well with a note as to which tree is preferred (sgi.com or kernel.org).

I'll update the wikis to bring it uptodate.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#9 lore raw
Quoted 6 lines · On Tuesday, 24 February 2009 19:46:24 Christoph Hellwig wrote:
Quoted 1 line · Mike Frysinger

but acl/attr have not. distros care about those packages as well.

Andreas who has done most of the maintaince already is taking these over formally, so they won't be released with the xfs tools anymore. I've put him on Cc so he can say when he expects to clear the backlog.

I think I have caught up with the backlog by now in my current trees:

http://git.kernel.org/?p=linux/kernel/git/agruen/attr.git
http://git.kernel.org/?p=linux/kernel/git/agruen/acl.git

The repos are based on Christoph's initial git import, with all additional fixes added that I'm aware of. Please let me know if anything is missing.

I expect that those repositories will move from where they are right now to their permanent new location soon, so please stay tuned ;-)

Thanks,
Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#10 lore raw
Quoted 10 lines · On Tuesday 24 February 2009 17:22:48 Andreas Gruenbacher wrote:
Quoted 6 lines · On Tuesday, 24 February 2009 19:46:24 Christoph Hellwig wrote:
Quoted 1 line · Mike Frysinger

but acl/attr have not. distros care about those packages as well.

Andreas who has done most of the maintaince already is taking these over formally, so they won't be released with the xfs tools anymore. I've put him on Cc so he can say when he expects to clear the backlog.

I think I have caught up with the backlog by now in my current trees:

Quoted 8 lines

http://git.kernel.org/?p=linux/kernel/git/agruen/attr.git
http://git.kernel.org/?p=linux/kernel/git/agruen/acl.git

The repos are based on Christoph's initial git import, with all additional fixes added that I'm aware of. Please let me know if anything is missing.

I expect that those repositories will move from where they are right now to their permanent new location soon, so please stay tuned ;-)

is this mailing list still going to be used for acl/attr ?

here are the patches i have for acl.git:
http://sources.gentoo.org/sys-apps/acl/files/acl-2.2.45-libtool.patch
http://sources.gentoo.org/sys-apps/acl/files/acl-2.2.45-linguas.patch

and for attr.git:
http://sources.gentoo.org/sys-apps/attr/files/attr-2.4.39-gettext.patch
http://sources.gentoo.org/sys-apps/attr/files/attr-2.4.43-linguas.patch

hopefully they should be self explanatory ... otherwise, feel free to ask of course
-mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#11 lore raw
Quoted 1 line · On Wednesday, 25 February 2009 0:45:25 Mike Frysinger wrote:

is this mailing list still going to be used for acl/attr ?

I would like to keep the discussion here until the packages have found a final new location, if SGI won't mind.

Quoted 2 lines

here are the patches i have for acl.git:
http://sources.gentoo.org/sys-apps/acl/files/acl-2.2.45-libtool.patch

Added.

Quoted 1 line

http://sources.gentoo.org/sys-apps/acl/files/acl-2.2.45-linguas.patch

Added.

Quoted 2 lines

and for attr.git:
http://sources.gentoo.org/sys-apps/attr/files/attr-2.4.39-gettext.patch

This functionality already exists (in both the acl and attr packages).

Quoted 1 line

http://sources.gentoo.org/sys-apps/attr/files/attr-2.4.43-linguas.patch

Added.

Thanks, Mike!

Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#12 lore raw
Quoted 5 lines · On Wednesday 25 February 2009 19:26:14 Andreas Gruenbacher wrote:
Quoted 2 lines · On Wednesday, 25 February 2009 0:45:25 Mike Frysinger wrote:

and for attr.git:
http://sources.gentoo.org/sys-apps/attr/files/attr-2.4.39-gettext.patch

This functionality already exists (in both the acl and attr packages).

was this in the last release (2.4.43) ? either way, can you highlight the code you're referring to so there's no confusion on my part ;) -mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#13 lore raw
Quoted 9 lines · On Thursday 26 February 2009 02:03:29 Mike Frysinger wrote:
Quoted 5 lines · On Wednesday 25 February 2009 19:26:14 Andreas Gruenbacher wrote:
Quoted 2 lines · On Wednesday, 25 February 2009 0:45:25 Mike Frysinger wrote:

and for attr.git:
http://sources.gentoo.org/sys-apps/attr/files/attr-2.4.39-gettext.patch

This functionality already exists (in both the acl and attr packages).

was this in the last release (2.4.43) ? either way, can you highlight the code you're referring to so there's no confusion on my part ;)

Hmm ... so there code that this patch adds to include/gettext.h already exists in include/config.h.in, but ENABLE_GETTEXT isn't being defined anywhere. So this part of your patch still seems to be needed, in both the attr and acl packages.

When I add it, I get this far:

	$ make aclocal.m4
	$ autoconf
	configure:3493: error: possibly undefined macro: AC_PROG_LIBTOOL
      If this token and others are legitimate, please use m4_pattern_allow.
      See the Autoconf documentation.

It turns out that aclocal.m4 is created with only the macros in the m4/ directory, and AC_PROG_LIBTOOL is not among them; this can be checked with:

	$ aclocal --acdir=m4 --verbose

This would usually be fixed by including the macros in m4/ in addition, but then I run into the next problem:

	$ aclocal -I m4
	$ autoconf
	[...]
	configure: error: cannot run /bin/sh ./config.sub

As per one of the libtool info pages,

Quoted 6 lines

In order to use libtool, you need to include the following files with your package:

`config.guess'
Attempt to guess a canonical system name.

Quoted 8 lines

`config.sub'
Canonical system name validation subroutine script.

`install-sh'
BSD-compatible `install' replacement script.

`ltmain.sh'
A generic script implementing basic libtool functionality.

after adding config.guess, config.sub from automake, and ltmain.sh from libtool (we already have our own install-sh), the package builds again.

I'm attaching a diff relative to the current repo excluding config.guess, config.sub, and ltmain.sh which should be on your systems from automake and libtool already.

This is quite insane. Is there a more reasonable way?

Thanks,
Andreas

.gitignore +1 -0
index a4d8b60..6391bdc 100644
@@ -25,3 +25,4 @@ attr-*.tar.gz
 *.o
 /po/attr.pot
 /setfattr/setfattr
+/libtool
Makefile +2 -1
index 7d2faad..e42a371 100644
@@ -11,6 +11,7 @@ endif
 
 CONFIGURE = configure include/builddefs include/config.h
 LSRCFILES = configure configure.in aclocal.m4 Makepkgs install-sh exports \
+	config.sub config.guess ltmain.sh \
 	README VERSION
 
 LDIRT = config.log .dep config.status config.cache confdefs.h conftest* \
@@ -63,7 +64,7 @@ include/config.h: include/builddefs
 	fi
 
 aclocal.m4::
-	aclocal --acdir=`pwd`/m4 --output=$@
+	aclocal -I m4
 
 install: default $(addsuffix -install,$(SUBDIRS))
 	$(INSTALL) -m 755 -d $(PKG_DOC_DIR)
m4/package_utilies.m4 +1 -0
index 6569e3b..38d623c 100644
@@ -100,6 +100,7 @@ AC_DEFUN([AC_PACKAGE_UTILITIES],
         xgettext=$XGETTEXT
         AC_SUBST(xgettext)
         AC_PACKAGE_NEED_UTILITY($1, "$xgettext", xgettext, gettext)
+	AC_DEFINE([ENABLE_GETTEXT], 1, [enable gettext])
     fi
 
     if test -z "$RPM"; then

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#14 lore raw
Quoted 55 lines · On Thursday 26 February 2009 07:23:08 Andreas Gruenbacher wrote:
Quoted 10 lines · On Thursday 26 February 2009 02:03:29 Mike Frysinger wrote:
Quoted 6 lines · On Wednesday 25 February 2009 19:26:14 Andreas Gruenbacher wrote:
Quoted 3 lines · On Wednesday, 25 February 2009 0:45:25 Mike Frysinger wrote:

and for attr.git:
http://sources.gentoo.org/sys-apps/attr/files/attr-2.4.39-gettext.pat
ch

This functionality already exists (in both the acl and attr packages).

was this in the last release (2.4.43) ? either way, can you highlight the code you're referring to so there's no confusion on my part ;)

Hmm ... so there code that this patch adds to include/gettext.h already exists in include/config.h.in, but ENABLE_GETTEXT isn't being defined anywhere. So this part of your patch still seems to be needed, in both the attr and acl packages.

When I add it, I get this far:

	$ make aclocal.m4
	$ autoconf
	configure:3493: error: possibly undefined macro: AC_PROG_LIBTOOL
      If this token and others are legitimate, please use m4_pattern_allow.
      See the Autoconf documentation.

It turns out that aclocal.m4 is created with only the macros in the m4/ directory, and AC_PROG_LIBTOOL is not among them; this can be checked with:

	$ aclocal --acdir=m4 --verbose

This would usually be fixed by including the macros in m4/ in addition, but then I run into the next problem:

	$ aclocal -I m4
	$ autoconf
	[...]
	configure: error: cannot run /bin/sh ./config.sub

As per one of the libtool info pages,

Quoted 6 lines

In order to use libtool, you need to include the following files with your package:

`config.guess'
Attempt to guess a canonical system name.

Quoted 8 lines

`config.sub'
Canonical system name validation subroutine script.

`install-sh'
BSD-compatible `install' replacement script.

`ltmain.sh'
A generic script implementing basic libtool functionality.

Quoted 8 lines

after adding config.guess, config.sub from automake, and ltmain.sh from libtool (we already have our own install-sh), the package builds again.

I'm attaching a diff relative to the current repo excluding config.guess, config.sub, and ltmain.sh which should be on your systems from automake and libtool already.

This is quite insane. Is there a more reasonable way?

yes. i dont know why the xfs progs have been packaging these autogenerated files by themselves.

- remove aclocal.m4 from git
- run aclocal -I m4
- run libtoolize -c -f
- run autoconf
- copy the remaining files libtool wanted (normally created by automake, but acl isnt using automake, so we cant use it to copy the files for us) -mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#15 lore raw
Quoted 7 lines · On Thursday, 26 February 2009 16:17:28 Mike Frysinger wrote:

yes. i dont know why the xfs progs have been packaging these autogenerated files by themselves.

- remove aclocal.m4 from git
- run aclocal -I m4
- run libtoolize -c -f
- run autoconf

You mean this should become the normal build process?

I was actually more thinking along the lines of reducing build dependencies by adding a few more generated files like ./configure ...

Quoted 2 lines

- copy the remaining files libtool wanted (normally created by automake, but acl isnt using automake, so we cant use it to copy the files for us)

Thanks,
Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#16 lore raw
Quoted 9 lines · On Thursday 26 February 2009 12:17:02 Andreas Gruenbacher wrote:
Quoted 7 lines · On Thursday, 26 February 2009 16:17:28 Mike Frysinger wrote:

yes. i dont know why the xfs progs have been packaging these autogenerated files by themselves.

- remove aclocal.m4 from git
- run aclocal -I m4
- run libtoolize -c -f
- run autoconf

Quoted 4 lines

You mean this should become the normal build process?

I was actually more thinking along the lines of reducing build dependencies by adding a few more generated files like ./configure ...

no, this is the process for going from live git to a dist tarball. people who download tarballs should never need autotools installed. people who build from live scms should.
-mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#17 lore raw
In reply to Andreas Gruenbacher
Quoted 12 lines · Andreas Gruenbacher
Quoted 8 lines · On Thursday, 26 February 2009 16:17:28 Mike Frysinger wrote:

yes. i dont know why the xfs progs have been packaging these autogenerated files by themselves.

- remove aclocal.m4 from git
- run aclocal -I m4
- run libtoolize -c -f
- run autoconf

Some packages provide a shell script called autogen.sh which automates all those tedious autotools steps. The Gnome autogen.sh is fairly general and does things like check for autotools versions and work out dynamically which tools to run.

http://svn.gnome.org/viewvc/gnome-common/trunk/autogen.sh?revision=3902&view=markup
http://svn.gnome.org/viewvc/gnome-common/trunk/macros2/gnome-autogen.sh?revision=3913&view=markup

You don't need all that of course, you could get away with a 5-line shell script. As long as you call it "autogen.sh" people should know what to do.

Quoted 6 lines

You mean this should become the normal build process?

I was actually more thinking along the lines of reducing build dependencies by adding a few more generated files like ./configure ...

Not in a version control system. Otherwise developers who do run autoconf, using a different version of autoconf from you, will get an enormous spurious diff in the configure script which they have to decide whether to ignore or not.

Signature · 8 lines
Greg Banks, P.Engineer, SGI Australian Software Group.
the brightly coloured sporks of revolution.
I don't speak for SGI.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#18 lore raw
Quoted 26 lines · On Friday, 27 February 2009 8:22:59 Greg Banks wrote:
Quoted 8 lines · Andreas Gruenbacher
Quoted 7 lines · On Thursday, 26 February 2009 16:17:28 Mike Frysinger wrote:

yes. i dont know why the xfs progs have been packaging these autogenerated files by themselves.

- remove aclocal.m4 from git
- run aclocal -I m4
- run libtoolize -c -f
- run autoconf

Some packages provide a shell script called autogen.sh which automates all those tedious autotools steps. The Gnome autogen.sh is fairly general and does things like check for autotools versions and work out dynamically which tools to run.

http://svn.gnome.org/viewvc/gnome-common/trunk/autogen.sh?revision=3902&vie
w=markup
http://svn.gnome.org/viewvc/gnome-common/trunk/macros2/gnome-autogen.sh?rev
ision=3913&view=markup

You don't need all that of course, you could get away with a 5-line shell script. As long as you call it "autogen.sh" people should know what to do.

Quoted 2 lines

You mean this should become the normal build process?

Quoted 7 lines
Quoted 2 lines

I was actually more thinking along the lines of reducing build dependencies by adding a few more generated files like ./configure ...

Not in a version control system. Otherwise developers who do run autoconf, using a different version of autoconf from you, will get an enormous spurious diff in the configure script which they have to decide whether to ignore or not.

Of course, the classic way is to keep all the generated files out of the repository. On the other hand, if we added all the generated files to the repository, all that would be needed would be creating a new release would be creating a tag and making a tarball. Have a look at cgit: if configured that way, the web interface offers to download any tagged version as a tarball. Those tarballs are generated on the fly (e.g.,
http://git.savannah.gnu.org/cgit/).

Jumping between different versions of the auto* tools would become quite ugly, and we would have to avoid jumping between versions. So maybe that's more trouble than it would be worth.

Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#19 lore raw
Quoted 40 lines · Andreas Gruenbacher
Quoted 29 lines · On Friday, 27 February 2009 8:22:59 Greg Banks wrote:
Quoted 10 lines · Andreas Gruenbacher
Quoted 8 lines · On Thursday, 26 February 2009 16:17:28 Mike Frysinger wrote:

yes. i dont know why the xfs progs have been packaging these autogenerated files by themselves.

- remove aclocal.m4 from git
- run aclocal -I m4
- run libtoolize -c -f
- run autoconf

Some packages provide a shell script called autogen.sh which automates all those tedious autotools steps. The Gnome autogen.sh is fairly general and does things like check for autotools versions and work out dynamically which tools to run.

http://svn.gnome.org/viewvc/gnome-common/trunk/autogen.sh?revision=3902&vie
w=markup
http://svn.gnome.org/viewvc/gnome-common/trunk/macros2/gnome-autogen.sh?rev
ision=3913&view=markup

You don't need all that of course, you could get away with a 5-line shell script. As long as you call it "autogen.sh" people should know what to do.

Quoted 2 lines

You mean this should become the normal build process?

Quoted 8 lines
Quoted 3 lines

I was actually more thinking along the lines of reducing build dependencies by adding a few more generated files like ./configure ...

Not in a version control system. Otherwise developers who do run autoconf, using a different version of autoconf from you, will get an enormous spurious diff in the configure script which they have to decide whether to ignore or not.

Quoted 4 lines

Of course, the classic way is to keep all the generated files out of the repository. On the other hand, if we added all the generated files to the repository, all that would be needed would be creating a new release would be creating a tag and making a tarball.

If the point here is to reduce the workload and potential errors involved in making a release, you could add a release: target to the Makefile to do the release procedure. That target only needs to work for the maintainer, so it's relatively easy. You could do the stored version bump, the tagging, the autotools futzing, and release tarball building in there.

Quoted 5 lines

Have a look at cgit: if configured that
way, the web interface offers to download any tagged version as a tarball. Those tarballs are generated on the fly (e.g.,
http://git.savannah.gnu.org/cgit/).

Oh, now I see where the idea of checking in generated files comes from :-)

Quoted 2 lines

Jumping between different versions of the auto* tools would become quite ugly,

Absolutely, you want to avoid that. But conversely you can't rely on everybody who ever needs to change configure.ac having exactly the same version of autoconf as the maintainer.

Signature · 8 lines
Greg Banks, P.Engineer, SGI Australian Software Group.
the brightly coloured sporks of revolution.
I don't speak for SGI.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#20 lore raw
Quoted 7 lines · On Saturday 28 February 2009 20:31:01 Greg Banks wrote:
Quoted 2 lines · Andreas Gruenbacher

Jumping between different versions of the auto* tools would become quite ugly,

Absolutely, you want to avoid that. But conversely you can't rely on everybody who ever needs to change configure.ac having exactly the same version of autoconf as the maintainer.

that isnt to say we should attempt to support every random version out there. if people want to make with autotools, it's reasonable to require them to have recent versions.
-mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#21 lore raw
In reply to Greg Banks
Quoted 6 lines · Greg Banks

If the point here is to reduce the workload and potential errors involved in making a release, you could add a release: target to the Makefile to do the release procedure. That target only needs to work for the maintainer, so it's relatively easy. You could do the stored version bump, the tagging, the autotools futzing, and release tarball building in there.

The xfs tools (and acl/attr which use the same buildsystem) have a Makepkgs script that's supposed to do this with some help from the built system.

I'm not sure we should add generating the tags and updating doc/CHANGES to it, but everything else is automated iff we actually use it.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#22 lore raw
In reply to Mike Frysinger
Quoted 7 lines · On Thursday, 26 February 2009 16:17:28 Mike Frysinger wrote:

[...] i dont know why the xfs progs have been packaging these autogenerated files by themselves.

- remove aclocal.m4 from git
- run aclocal -I m4
- run libtoolize -c -f
- run autoconf

Okay, I've done that now in the configure Makefile target. This is an improvement we definitely want, independent of whether or not we end up shipping generated files.

Thanks,
Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#23 lore raw
Quoted 3 lines · Andreas Gruenbacher

Okay, I've done that now in the configure Makefile target. This is an improvement we definitely want, independent of whether or not we end up shipping generated files.

I've tried to port the patch you checked into xfsprogs (see below for the diff), but it fails to build for me. Any idea what might have gone wrong?

Makefile +6 -6
index 133e496..4b61e41 100644
@@ -12,8 +12,9 @@ endif
 CONFIGURE = configure include/builddefs include/platform_defs.h
 LSRCFILES = configure configure.in Makepkgs aclocal.m4 install-sh README VERSION
 
-LDIRT = config.log .dep config.status config.cache confdefs.h conftest* \
-	Logs/* built .census install.* install-dev.* *.gz
+LDIRT = config.log config.status config.cache config.guess config.sub \
+	confdefs.h ltmain.sh libtool built .census \
+	Logs/* conftest* install.* install-dev.* *.dep *.gz
 
 LIB_SUBDIRS = libxfs libxlog libxcmd libhandle libdisk
 TOOL_SUBDIRS = copy db estimate fsck fsr growfs io logprint mkfs quota \
@@ -21,7 +22,7 @@ TOOL_SUBDIRS = copy db estimate fsck fsr growfs io logprint mkfs quota \
 
 SUBDIRS = include $(LIB_SUBDIRS) $(TOOL_SUBDIRS)
 
-default: include/builddefs include/platform_defs.h
+default: configure include/builddefs include/platform_defs.h
 ifeq ($(HAVE_BUILDDEFS), no)
 	$(MAKE) -C . $@
 else
@@ -46,6 +47,8 @@ clean:	# if configure hasn't run, nothing to clean
 endif
 
 configure include/builddefs:
+	libtoolize -c -f
+	aclocal -I m4
 	autoconf
 	./configure \
 		--prefix=/ \
@@ -68,9 +71,6 @@ include/platform_defs.h: include/builddefs
 		$(MAKE) $(AM_MAKEFLAGS) include/builddefs; \
 	fi
 
-aclocal.m4::
-	aclocal --acdir=`pwd`/m4 --output=$@
-
 install: default $(addsuffix -install,$(SUBDIRS))
 	$(INSTALL) -m 755 -d $(PKG_DOC_DIR)
 	$(INSTALL) -m 644 README $(PKG_DOC_DIR)
configure.in +3 -0
index 4e4e50c..531d7d0 100644
@@ -1,6 +1,9 @@
 AC_INIT(include/libxfs.h)
+AC_CONFIG_MACRO_DIR([m4])
 AC_CONFIG_HEADER(include/platform_defs.h)
 
+AC_PROG_LIBTOOL
+
 AC_ARG_ENABLE(shared,
 [ --enable-shared=[yes/no] Enable use of shared libraries [default=yes]],,
 	enable_shared=yes)

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#24 lore raw
Quoted 3 lines · On Wednesday, 4 March 2009 18:27:27 Christoph Hellwig wrote:

I've tried to port the patch you checked into xfsprogs (see below for the diff), but it fails to build for me. Any idea what might have gone wrong?

git://git.kernel.org/pub/scm/fs/xfs/xfsprogs-dev.git is missing the following commit from ssh://master.kernel.org/pub/scm/linux/kernel/git/agruen/acl.git:

commit 8de7788c29d2656f8d52d98327a80e4dfd1e3898
Author: Barry Naujok <[email protected]>
Date: Tue Jan 23 14:51:14 2007 +0000

Fix cross-compile issues with libtool and compiler.

This is equivalent to commit de7b3f6 from Barry Naujok <[email protected]> in the acl package, and part of the Gentoo package, as pointed out by Mike Frysinger <[email protected]>.

While looking into this, I noticed that the attr and acl packages were calling AC_PROG_LIBTOOL both in configure.in and m4/package_utilies.m4. I have removed the redundant call from m4/package_utilies.m4 now.

Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#25 lore raw
Quoted 19 lines · On Sunday 08 March 2009 14:09:28 Andreas Gruenbacher wrote:
Quoted 3 lines · On Wednesday, 4 March 2009 18:27:27 Christoph Hellwig wrote:

I've tried to port the patch you checked into xfsprogs (see below for the diff), but it fails to build for me. Any idea what might have gone wrong?

git://git.kernel.org/pub/scm/fs/xfs/xfsprogs-dev.git is missing the following commit from ssh://master.kernel.org/pub/scm/linux/kernel/git/agruen/acl.git:

commit 8de7788c29d2656f8d52d98327a80e4dfd1e3898
Author: Barry Naujok <[email protected]>
Date: Tue Jan 23 14:51:14 2007 +0000

Fix cross-compile issues with libtool and compiler.

This is equivalent to commit de7b3f6 from Barry Naujok
<[email protected]> in the acl package, and part of the Gentoo package, as pointed out by Mike Frysinger <[email protected]>.

Quoted 5 lines

While looking into this, I noticed that the attr and acl packages were calling AC_PROG_LIBTOOL both in configure.in and m4/package_utilies.m4. I have removed the redundant call from m4/package_utilies.m4 now.

As it turns out, I was running into more trouble with libtoolize from libtool 2.2.6. See this commit in attr.git:

commit 604290f79a199eb0c73a0cd05a473e1801a00673
Author: Andreas Gruenbacher <[email protected]>
Date: Tue Mar 10 17:00:35 2009 +0100

More libtoolize fixes

Recent versions of libtool behave slightly differently, which causes some breakage in how libtoolize was used here. Make sure that libtoolize adds the auxiliary files (config.guess and config.sub).

Move install-sh into include/ so that libtoolize does not destroy it.

Split up the ``make clean'' and ``make distclean'' targets: the former removes all files generated during a build. The latter removes all files generated by libtoolize, autoconf, and configure as well.

With these changes, the build/*.src.tar.gz tarball should now build properly without depending on the version of libtool on the system with:

./configure
make
make install

(The acl.git repository has the equivalent changes.)

Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#26 lore raw
Quoted 8 lines · Andreas Gruenbacher

As it turns out, I was running into more trouble with libtoolize from libtool 2.2.6. See this commit in attr.git:

commit 604290f79a199eb0c73a0cd05a473e1801a00673
Author: Andreas Gruenbacher <[email protected]>
Date: Tue Mar 10 17:00:35 2009 +0100

More libtoolize fixes

That patch breaks the commit for me:

hch@brick:~/work/attr$ make
libtoolize -c -i -f
libtoolize: unrecognized option `-i'
Try `libtoolize --help' for more information.
make: *** [include/builddefs] Error 1

This is on debian -testing with libtool 1.5.26

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#27 lore raw
Quoted 10 lines · On Monday 16 March 2009 03:05:50 Christoph Hellwig wrote:
Quoted 8 lines · Andreas Gruenbacher

As it turns out, I was running into more trouble with libtoolize from libtool 2.2.6. See this commit in attr.git:

commit 604290f79a199eb0c73a0cd05a473e1801a00673
Author: Andreas Gruenbacher <[email protected]>
Date: Tue Mar 10 17:00:35 2009 +0100

More libtoolize fixes

Quoted 7 lines

That patch breaks the commit for me:

hch@brick:~/work/attr$ make
libtoolize -c -i -f
libtoolize: unrecognized option `-i'
Try `libtoolize --help' for more information.
make: *** [include/builddefs] Error 1

-i is new to libtool-2

usually what i do is something like:
args=
libtoolize -n --install 2>/dev/null && args="-i"
libtoolize -c -f $args
-mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#28 lore raw
Quoted 4 lines · On Monday, 16 March 2009 8:31:48 Mike Frysinger wrote:

usually what i do is something like:
args=
libtoolize -n --install 2>/dev/null && args="-i"
libtoolize -c -f $args

Sigh, the usual GNU utility problem. I've added a compatibility fix to attr and acl now. Thanks!

Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#29 lore raw
Quoted 8 lines · Andreas Gruenbacher
Quoted 4 lines · On Monday, 16 March 2009 8:31:48 Mike Frysinger wrote:

usually what i do is something like:
args=
libtoolize -n --install 2>/dev/null && args="-i"
libtoolize -c -f $args

Sigh, the usual GNU utility problem. I've added a compatibility fix to attr and acl now. Thanks!

Ok. Can anyone review the cumulated autotools/libtool patch for xfsprogs below?

Subject: Automake and libtool fixes
From: Andreas Gruenbacher <[email protected]>

Remove aclocal.m4 from the repository and generate it when needed. Move the AC_PROG_LIBTOOL autoconf macro and use libtoolize according to the libtool info pages.

Make sure that libtoolize adds the auxiliary files (config.guess and config.sub).

Move install-sh into include/ so that libtoolize does not destroy it.

Split up the ``make clean'' and ``make distclean'' targets: the former removes all files generated during a build. The latter removes all files generated by libtoolize, autoconf, and configure as well.

Makefile +21 -8
Index: xfsprogs-dev/Makefile
===================================================================
@@ -9,11 +9,18 @@ ifeq ($(HAVE_BUILDDEFS), yes)
 include $(TOPDIR)/include/builddefs
 endif
 
-CONFIGURE = configure include/builddefs include/platform_defs.h
-LSRCFILES = configure configure.in Makepkgs aclocal.m4 install-sh README VERSION
+CONFIGURE = \
+	aclocal.m4 \
+	configure config.guess config.sub \
+	ltmain.sh m4/libtool.m4 m4/ltoptions.m4 m4/ltsugar.m4 \
+	m4/ltversion.m4 m4/lt~obsolete.m4 \
+	include/builddefs include/platform_defs.h
+LSRCFILES = \
+	configure.in Makepkgs install-sh README VERSION \
+	configure aclocal.m4
 
-LDIRT = config.log .dep config.status config.cache confdefs.h conftest* \
-	Logs/* built .census install.* install-dev.* *.gz
+LDIRT = config.log config.status config.cache confdefs.h built .census \
+	Logs/* conftest* install.* install-dev.* *.dep *.gz
 
 LIB_SUBDIRS = libxfs libxlog libxcmd libhandle libdisk
 TOOL_SUBDIRS = copy db estimate fsck fsr growfs io logprint mkfs quota \
@@ -21,7 +28,7 @@ TOOL_SUBDIRS = copy db estimate fsck fsr
 
 SUBDIRS = include $(LIB_SUBDIRS) $(TOOL_SUBDIRS)
 
-default: include/builddefs include/platform_defs.h
+default: configure include/builddefs include/platform_defs.h
 ifeq ($(HAVE_BUILDDEFS), no)
 	$(MAKE) -C . $@
 else
@@ -45,7 +52,15 @@ else
 clean:	# if configure hasn't run, nothing to clean
 endif
 
+# Recent versions of libtool require the -i option for copying auxiliary
+# files (config.sub, config.guess, install-sh, ltmain.sh), while older
+# versions will copy those files anyway, and don't understand -i.
+LIBTOOLIZE_INSTALL = `libtoolize -n -i >/dev/null 2>/dev/null && echo -i`
+
 configure include/builddefs:
+	libtoolize -c $(LIBTOOLIZE_INSTALL) -f
+	cp include/install-sh .
+	aclocal -I m4
 	autoconf
 	./configure \
 		--prefix=/ \
@@ -68,9 +83,6 @@ include/platform_defs.h: include/buildde
 		$(MAKE) $(AM_MAKEFLAGS) include/builddefs; \
 	fi
 
-aclocal.m4::
-	aclocal --acdir=`pwd`/m4 --output=$@
-
 install: default $(addsuffix -install,$(SUBDIRS))
 	$(INSTALL) -m 755 -d $(PKG_DOC_DIR)
 	$(INSTALL) -m 644 README $(PKG_DOC_DIR)
@@ -90,4 +102,5 @@ install-qa: install $(addsuffix -install
 
 realclean distclean: clean
 	rm -f $(LDIRT) $(CONFIGURE)
+	rm -f include/builddefs include/config.h install-sh libtool
 	rm -rf autom4te.cache Logs
configure.in +4 -0
Index: xfsprogs-dev/configure.in
===================================================================
@@ -1,6 +1,10 @@
 AC_INIT(include/libxfs.h)
+AC_CONFIG_AUX_DIR([.])
+AC_CONFIG_MACRO_DIR([m4])
 AC_CONFIG_HEADER(include/platform_defs.h)
 
+AC_PROG_LIBTOOL
+
 AC_ARG_ENABLE(shared,
 [ --enable-shared=[yes/no] Enable use of shared libraries [default=yes]],,
 	enable_shared=yes)
m4/package_utilies.m4 +0 -2
Index: xfsprogs-dev/m4/package_utilies.m4
===================================================================
@@ -32,8 +32,6 @@ AC_DEFUN([AC_PACKAGE_UTILITIES],
     AC_SUBST(make)
     AC_PACKAGE_NEED_UTILITY($1, "$make", make, [GNU make])
 
-    AC_PROG_LIBTOOL
-
     if test -z "$TAR"; then
         AC_PATH_PROG(TAR, tar,, /usr/freeware/bin:/bin:/usr/local/bin:/usr/bin)
     fi
include/Makefile +2 -2
Index: xfsprogs-dev/include/Makefile
===================================================================
@@ -36,8 +36,8 @@ HFILES += $(PKG_PLATFORM).h
 PHFILES = darwin.h freebsd.h irix.h linux.h
 DKHFILES = volume.h fstyp.h dvh.h
 LSRCFILES = $(shell echo $(PHFILES) | sed -e "s/$(PKG_PLATFORM).h//g")
-LSRCFILES += platform_defs.h.in builddefs.in buildmacros buildrules $(DKHFILES)
-LSRCFILES += command.h input.h path.h project.h unaligned.h
+LSRCFILES += platform_defs.h.in builddefs.in buildmacros buildrules install-sh
+LSRCFILES += $(DKHFILES) command.h input.h path.h project.h unaligned.h
 LDIRT = xfs disk
 
 default install: xfs disk
include/install-sh new file +349 -0

349 lines not on this page · raw message

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#30 lore raw

Makefile ifeq ($(HAVE_BUILDDEFS), yes) · line 16

Quoted 2 lines
Makefile +2 -0
+	ltmain.sh m4/libtool.m4 m4/ltoptions.m4 m4/ltsugar.m4 \
+	m4/ltversion.m4 m4/lt~obsolete.m4 \

Btw, where do these m4 files come from? Is this another libtool version dependency? I can't find them anywhere for either acl/attr or the patched xfsprogs

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#31 lore raw
Quoted 8 lines · On Tuesday, 24 March 2009 14:12:49 Christoph Hellwig wrote:

Makefile ifeq ($(HAVE_BUILDDEFS), yes) · line 16

Quoted 2 lines
Makefile +2 -0
+	ltmain.sh m4/libtool.m4 m4/ltoptions.m4 m4/ltsugar.m4 \
+	m4/ltversion.m4 m4/lt~obsolete.m4 \

Btw, where do these m4 files come from?

From libtool via libtoolize. The AC_PROG_LIBTOOL macro pulls them in AFAICT.

Is this another libtool version dependency?

It would seem so but the build system simply ignores files in the file list that don't exist. This is all rather ugly, but at least the result works...

Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#32 lore raw
Quoted 2 lines · Andreas Gruenbacher

It would seem so but the build system simply ignores files in the file list that don't exist. This is all rather ugly, but at least the result works...

Oh well. I've now pushed these patches to xfsprogs, xfsdump and dmapi as authored by you and reviewed by me as they were straight-forward ports.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#33 lore raw
In reply to Andreas Gruenbacher
Quoted 14 lines · Andreas Gruenbacher
Quoted 3 lines · On Wednesday, 4 March 2009 18:27:27 Christoph Hellwig wrote:

I've tried to port the patch you checked into xfsprogs (see below for the diff), but it fails to build for me. Any idea what might have gone wrong?

git://git.kernel.org/pub/scm/fs/xfs/xfsprogs-dev.git is missing the following commit from ssh://master.kernel.org/pub/scm/linux/kernel/git/agruen/acl.git:

commit 8de7788c29d2656f8d52d98327a80e4dfd1e3898
Author: Barry Naujok <[email protected]>
Date: Tue Jan 23 14:51:14 2007 +0000

Fix cross-compile issues with libtool and compiler.

Quoted 7 lines

This is equivalent to commit de7b3f6 from Barry Naujok <[email protected]> in the acl package, and part of the Gentoo package, as pointed out by Mike Frysinger <[email protected]>.

While looking into this, I noticed that the attr and acl packages were calling

Thank, I've pushed this patch to the xfsprogs, xfsdump and dmapi -dev repositories.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#34 lore raw
In reply to Andreas Gruenbacher

I did some make dist and build the tarball tests with acl,attr,xfsprogs & co.

We still run libtoolize, aclocal and autoconf when building a clean tree. I wonder if we should just get rid of the whole configure target and require a manual ./configure like most tools?

Nathan, any comments on that from the Debian perspective?

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#35 lore raw
Quoted 6 lines · On Friday 27 March 2009 12:02:04 Christoph Hellwig wrote:

I did some make dist and build the tarball tests with acl,attr,xfsprogs & co.

We still run libtoolize, aclocal and autoconf when building a clean tree. I wonder if we should just get rid of the whole configure target and require a manual ./configure like most tools?

make dist should produce a tarball that, when unpacked, people can run:
./configure && make && make install
-mike

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#36 lore raw
Quoted 10 lines · Mike Frysinger
Quoted 6 lines · On Friday 27 March 2009 12:02:04 Christoph Hellwig wrote:

I did some make dist and build the tarball tests with acl,attr,xfsprogs & co.

We still run libtoolize, aclocal and autoconf when building a clean tree. I wonder if we should just get rid of the whole configure target and require a manual ./configure like most tools?

make dist should produce a tarball that, when unpacked, people can run:
./configure && make && make install

Yes. The added complication is that the various xfs-cmds derived tools also run ./configure from a default target make which complicates the build system.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#37 lore raw
In reply to Andreas Gruenbacher
Quoted 8 lines · Andreas Gruenbacher
Quoted 6 lines · On Tuesday, 24 February 2009 19:46:24 Christoph Hellwig wrote:
Quoted 1 line · Mike Frysinger

but acl/attr have not. distros care about those packages as well.

Andreas who has done most of the maintaince already is taking these over formally, so they won't be released with the xfs tools anymore. I've put him on Cc so he can say when he expects to clear the backlog.

Quoted 7 lines

I think I have caught up with the backlog by now in my current trees:

http://git.kernel.org/?p=linux/kernel/git/agruen/attr.git
http://git.kernel.org/?p=linux/kernel/git/agruen/acl.git

The repos are based on Christoph's initial git import, with all additional fixes added that I'm aware of. Please let me know if anything is missing.

Do you plan to take over nfs4acl, too?

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#38 lore raw
Quoted 1 line · On Friday, 27 February 2009 0:03:00 Christoph Hellwig wrote:

Do you plan to take over nfs4acl, too?

I can sure copy the repository over from you if that is any help. How that project will proceed is still unclear though, so don't expect a lot to happen there anytime soon.

Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#39 lore raw
Quoted 8 lines · Andreas Gruenbacher
Quoted 1 line · On Friday, 27 February 2009 0:03:00 Christoph Hellwig wrote:

Do you plan to take over nfs4acl, too?

I can sure copy the repository over from you if that is any help. How that project will proceed is still unclear though, so don't expect a lot to happen there anytime soon.

Andreas

I'd check in w/ Greg on that too, maybe?

-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#40 lore raw
Quoted 6 lines · Eric Sandeen
Quoted 4 lines · Andreas Gruenbacher
Quoted 2 lines · On Friday, 27 February 2009 0:03:00 Christoph Hellwig wrote:

Do you plan to take over nfs4acl, too?

Quoted 11 lines
Quoted 6 lines

I can sure copy the repository over from you if that is any help. How that project will proceed is still unclear though, so don't expect a lot to happen there anytime soon.

Andreas

I'd check in w/ Greg on that too, maybe?

I think the best thing for nfs4acl is for there to be repo owned by Andreas and adjacent to the attr and acl ones.

I'd also like eventually to see a sensible and unified nfs4acl kernel repo so we can fix up the mess that is the current crop of nfs4acl kernel trees on oss.sgi.com. For various reasons I'd prefer to see that on kernel.org or linux-nfs.org rather than oss.sgi.com. However I doubt I'll have any chance to do anything on that in the next few weeks.

Signature · 8 lines
Greg Banks, P.Engineer, SGI Australian Software Group.
the brightly coloured sporks of revolution.
I don't speak for SGI.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#41 lore raw
In reply to Eric Sandeen
Quoted 11 lines · On Saturday, 28 February 2009 23:27:28 Eric Sandeen wrote:
Quoted 8 lines · Andreas Gruenbacher
Quoted 1 line · On Friday, 27 February 2009 0:03:00 Christoph Hellwig wrote:

Do you plan to take over nfs4acl, too?

I can sure copy the repository over from you if that is any help. How that project will proceed is still unclear though, so don't expect a lot to happen there anytime soon.

Andreas

I'd check in w/ Greg on that too, maybe?

Yes, makes sense. I've copied Christoph's tree over and set up permissions; you should have access now.

Greg doesn't have a kernel.org account yet, so I can't give him access at the moment.

git://git.kernel.org/pub/scm/linux/kernel/git/agruen/nfs4acl.git
ssh://master.kernel.org/pub/scm/linux/kernel/git/agruen/nfs4acl.git

Would you like write access to the attr.git and acl.git trees as well?

Thanks,
Andreas

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#42 lore raw
Quoted 10 lines · Andreas Gruenbacher

Yes, makes sense. I've copied Christoph's tree over and set up permissions; you should have access now.

Greg doesn't have a kernel.org account yet, so I can't give him access at the moment.

git://git.kernel.org/pub/scm/linux/kernel/git/agruen/nfs4acl.git
ssh://master.kernel.org/pub/scm/linux/kernel/git/agruen/nfs4acl.git

Would you like write access to the attr.git and acl.git trees as well?

Greg or me? Don't think I want to do too much on it. I'll just bounce you whatever build system updates seem like fitting for acl/attr too. Still a few I need to send you.

But Nathan Scott might want it as he wants to keep the debian packaging in the upstream git, and I suspect acl/attr aren't different from the xfs tools in that respect.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs