[arch-dev-public] [signoff] devtools-0.8.0
Hi, A new devtools release has been pushed to [testing]. Everything works on a brief test here, but I thought it a good idea to do some testing before pushing to [extra]. Allan
On Sun, Nov 1, 2009 at 12:32 AM, Allan McRae <allan@archlinux.org> wrote:
Hi,
A new devtools release has been pushed to [testing]. Everything works on a brief test here, but I thought it a good idea to do some testing before pushing to [extra].
Allan
I'm having a problem with makechrootpkg. In this commit: http://projects.archlinux.org/devtools.git/commit/?id=71a6d2c89587da5299b5a6... The line: source $uniondir/etc/makepkg.conf was removed. This cause the package sources to be downloaded even if they are in my local source cache. Was it removed by mistake? If it is intended, then how does makechrootpkg's SRCDEST (and PKGDEST) variables get assigned? Did the configuration changed? Eric
Eric Bélanger wrote:
On Sun, Nov 1, 2009 at 12:32 AM, Allan McRae <allan@archlinux.org> wrote:
Hi,
A new devtools release has been pushed to [testing]. Everything works on a brief test here, but I thought it a good idea to do some testing before pushing to [extra].
Allan
I'm having a problem with makechrootpkg. In this commit: http://projects.archlinux.org/devtools.git/commit/?id=71a6d2c89587da5299b5a6...
The line: source $uniondir/etc/makepkg.conf
was removed. This cause the package sources to be downloaded even if they are in my local source cache. Was it removed by mistake? If it is intended, then how does makechrootpkg's SRCDEST (and PKGDEST) variables get assigned? Did the configuration changed?
That does look like a bad "fix". From memory, the whole issue here is for people who use makechrootpkg to build i686 and x86_64 packages on the one machine. Sourcing makepkg.conf on the host machine is bad in that case as it might not be for the correct architecture. So, the sourcing on makepkg.conf was moved to after the build so that the chroots makepkg.conf is used at all times (WARNING: set your PACKAGER variable there...). Now the host makepkg.conf is only sourced get the SRCDEST and PKGDEST to move new sources/packages at the end of the build. This obviously breaks copying source files from your SRCDEST as that is not sourced beforehand. I dislike this change (as I am too lazy to set MAKEFLAGS and PACKAGER in all my chroots) and so have been working with a makechrootpkg with that whole change reverted. I then have a "makechrootpkg64" script that sources /etc/makepkg64.conf instead... Reverting those changes a providing a flag for which makepkg config file to use may be a fix for this whole issue. Allan
On Mon, Nov 2, 2009 at 4:02 AM, Allan McRae <allan@archlinux.org> wrote:
Eric Bélanger wrote:
On Sun, Nov 1, 2009 at 12:32 AM, Allan McRae <allan@archlinux.org> wrote:
Hi,
A new devtools release has been pushed to [testing]. Everything works on a brief test here, but I thought it a good idea to do some testing before pushing to [extra].
Allan
I'm having a problem with makechrootpkg. In this commit:
http://projects.archlinux.org/devtools.git/commit/?id=71a6d2c89587da5299b5a6...
The line: source $uniondir/etc/makepkg.conf
was removed. This cause the package sources to be downloaded even if they are in my local source cache. Was it removed by mistake? If it is intended, then how does makechrootpkg's SRCDEST (and PKGDEST) variables get assigned? Did the configuration changed?
That does look like a bad "fix". From memory, the whole issue here is for people who use makechrootpkg to build i686 and x86_64 packages on the one machine. Sourcing makepkg.conf on the host machine is bad in that case as it might not be for the correct architecture. So, the sourcing on makepkg.conf was moved to after the build so that the chroots makepkg.conf is used at all times (WARNING: set your PACKAGER variable there...). Now the host makepkg.conf is only sourced get the SRCDEST and PKGDEST to move new sources/packages at the end of the build. This obviously breaks copying source files from your SRCDEST as that is not sourced beforehand.
I dislike this change (as I am too lazy to set MAKEFLAGS and PACKAGER in all my chroots) and so have been working with a makechrootpkg with that whole change reverted. I then have a "makechrootpkg64" script that sources /etc/makepkg64.conf instead... Reverting those changes a providing a flag for which makepkg config file to use may be a fix for this whole issue.
Allan
Another way is to get the SRCDEST and PKGDEST without sourcing the makepkg.conf: eval $(grep SRCDEST /etc/makepkg.conf) eval $(grep PKGDEST /etc/makepkg.conf)
On Mon, Nov 2, 2009 at 06:45, Eric Bélanger <snowmaniscool@gmail.com> wrote:
Another way is to get the SRCDEST and PKGDEST without sourcing the makepkg.conf:
eval $(grep SRCDEST /etc/makepkg.conf) eval $(grep PKGDEST /etc/makepkg.conf)
This would have to be eval $(grep '^SRCDEST=' /etc/makepkg.conf so that it wouldn't get any commented lines.
On Mon, Nov 2, 2009 at 8:08 AM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
On Mon, Nov 2, 2009 at 06:45, Eric Bélanger <snowmaniscool@gmail.com> wrote:
Another way is to get the SRCDEST and PKGDEST without sourcing the makepkg.conf:
eval $(grep SRCDEST /etc/makepkg.conf) eval $(grep PKGDEST /etc/makepkg.conf)
This would have to be eval $(grep '^SRCDEST=' /etc/makepkg.conf so that it wouldn't get any commented lines.
I'm pretty sure eval-ing a commented line won't do a whole lot: $ eval '#SRCDEST=foobar' $ echo $SRCDEST $ eval 'SRCDEST=foobar' $ echo $SRCDEST foobar
On Mon, Nov 2, 2009 at 10:12, Dan McGee <dpmcgee@gmail.com> wrote:
I'm pretty sure eval-ing a commented line won't do a whole lot: $ eval '#SRCDEST=foobar' $ echo $SRCDEST
$ eval 'SRCDEST=foobar' $ echo $SRCDEST foobar
You know... That is a pretty good point. >.> I blame it being too early in the morning :P
On Mon, Nov 2, 2009 at 9:17 AM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
On Mon, Nov 2, 2009 at 10:12, Dan McGee <dpmcgee@gmail.com> wrote:
I'm pretty sure eval-ing a commented line won't do a whole lot: $ eval '#SRCDEST=foobar' $ echo $SRCDEST
$ eval 'SRCDEST=foobar' $ echo $SRCDEST foobar
You know... That is a pretty good point. >.>
I blame it being too early in the morning :P
Haha no worries, I had just already had my first cup of coffee. :) -Dan
On Mon, Nov 2, 2009 at 10:31 AM, Dan McGee <dpmcgee@gmail.com> wrote:
On Mon, Nov 2, 2009 at 9:17 AM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
On Mon, Nov 2, 2009 at 10:12, Dan McGee <dpmcgee@gmail.com> wrote:
I'm pretty sure eval-ing a commented line won't do a whole lot: $ eval '#SRCDEST=foobar' $ echo $SRCDEST
$ eval 'SRCDEST=foobar' $ echo $SRCDEST foobar
You know... That is a pretty good point. >.>
I blame it being too early in the morning :P
Haha no worries, I had just already had my first cup of coffee. :)
-Dan
Actually, Daenyth's method is better than mine. If you have something like this in your makepkg.conf: #SRCDEST=/mnt/hd-ext SRCDEST=/var/cache/pacman/src my eval will fail but Daenyth's eval with regex will work. Eric
On Mon, Nov 2, 2009 at 5:55 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
On Mon, Nov 2, 2009 at 10:31 AM, Dan McGee <dpmcgee@gmail.com> wrote:
On Mon, Nov 2, 2009 at 9:17 AM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
On Mon, Nov 2, 2009 at 10:12, Dan McGee <dpmcgee@gmail.com> wrote:
I'm pretty sure eval-ing a commented line won't do a whole lot: $ eval '#SRCDEST=foobar' $ echo $SRCDEST
$ eval 'SRCDEST=foobar' $ echo $SRCDEST foobar
You know... That is a pretty good point. >.>
I blame it being too early in the morning :P
Haha no worries, I had just already had my first cup of coffee. :)
-Dan
Actually, Daenyth's method is better than mine. If you have something like this in your makepkg.conf:
#SRCDEST=/mnt/hd-ext SRCDEST=/var/cache/pacman/src
my eval will fail but Daenyth's eval with regex will work.
Eric
I just sent a patch (seperate thread in this ML). Eric
Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> --- makechrootpkg | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/makechrootpkg b/makechrootpkg index 5095425..d1dcf32 100755 --- a/makechrootpkg +++ b/makechrootpkg @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then rm -rf "$uniondir/build/"* fi +eval $(grep '^SRCDEST=' /etc/makepkg.conf) +eval $(grep '^PKGDEST=' /etc/makepkg.conf) + [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then echo "Setting PKGDEST in makepkg.conf" -- 1.6.5.2
On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> --- makechrootpkg | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/makechrootpkg b/makechrootpkg index 5095425..d1dcf32 100755 --- a/makechrootpkg +++ b/makechrootpkg @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then rm -rf "$uniondir/build/"* fi
+eval $(grep '^SRCDEST=' /etc/makepkg.conf) +eval $(grep '^PKGDEST=' /etc/makepkg.conf) + [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then echo "Setting PKGDEST in makepkg.conf"
The eval seems slightly dangerous to me... does anyone else have this concern, or am I being too careful?
On Thu, Nov 5, 2009 at 12:05, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
The eval seems slightly dangerous to me... does anyone else have this concern, or am I being too careful?
eval is always dangerous. In this case, however, it's eval-ing from a text file only writable by root. If an attacker has root write permissions, you have more to worry about than this.
On Thu, Nov 5, 2009 at 10:38 AM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
On Thu, Nov 5, 2009 at 12:05, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
The eval seems slightly dangerous to me... does anyone else have this concern, or am I being too careful?
eval is always dangerous. In this case, however, it's eval-ing from a text file only writable by root. If an attacker has root write permissions, you have more to worry about than this.
That's a fair enough point. I was wondering, though, if it might be more prudent to sed out the value and actually set it with no eval at all. Does anyone actually use inline execution in their makepkg.conf?
On Thu, Nov 5, 2009 at 12:41, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
That's a fair enough point. I was wondering, though, if it might be more prudent to sed out the value and actually set it with no eval at all. Does anyone actually use inline execution in their makepkg.conf?
Even if they do, should we support it? I don't think it makes a ton of difference honestly. Perhaps another option would be to give mkarchroot some flags to define the variables on chroot creation?
Daenyth Blank wrote:
On Thu, Nov 5, 2009 at 12:05, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
The eval seems slightly dangerous to me... does anyone else have this concern, or am I being too careful?
eval is always dangerous. In this case, however, it's eval-ing from a text file only writable by root. If an attacker has root write permissions, you have more to worry about than this.
True, but I still prefer to be extra careful, as /etc/makepkg.conf might have been compromised through other channels.
Aaron Griffin wrote:
On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> --- makechrootpkg | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/makechrootpkg b/makechrootpkg index 5095425..d1dcf32 100755 --- a/makechrootpkg +++ b/makechrootpkg @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then rm -rf "$uniondir/build/"* fi
+eval $(grep '^SRCDEST=' /etc/makepkg.conf) +eval $(grep '^PKGDEST=' /etc/makepkg.conf) + [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then echo "Setting PKGDEST in makepkg.conf"
The eval seems slightly dangerous to me... does anyone else have this concern, or am I being too careful?
I agree. It is more than "slightly" dangerous. If makepkg.conf contained a line such as: PKGDEST="blabla" && rm -rf / then the eval would indeed execute "rm -rf /". This is particularly bad since makechrootpkg is called with sudo ... Better look for a safer alternative. What about this? PKGDEST=$(source /etc/makepkg.conf && echo $PKGDEST) SRCDEST=$(source /etc/makepkg.conf && echo $SRCDEST) This is not very pretty as we need to source makepkg.conf twice, but at least it is safer. We could also do something like: ORIGDESTDIRS=$(source /etc/makepkg.conf && echo $PKGDEST $SRCDEST) PKGDEST=$(echo $ORIGDESTDIRS | cut -d' ' -f1) SRCDEST=$(echo $ORIGDESTDIRS | cut -d' ' -f2)
On Thu, Nov 5, 2009 at 12:53, Firmicus <Firmicus@gmx.net> wrote:
Better look for a safer alternative. What about this?
PKGDEST=$(source /etc/makepkg.conf && echo $PKGDEST) That's actually more dangerous than the eval, since you are essentially eval-ing the entire file.
Daenyth Blank wrote:
On Thu, Nov 5, 2009 at 12:53, Firmicus <Firmicus@gmx.net> wrote:
Better look for a safer alternative. What about this?
PKGDEST=$(source /etc/makepkg.conf && echo $PKGDEST)
That's actually more dangerous than the eval, since you are essentially eval-ing the entire file.
Again you're right. Somehow I thought source only affected variables, but actually it does also eval each line of the sourced file. Sorry for the noise.
On Thu, Nov 5, 2009 at 10:53 AM, Firmicus <Firmicus@gmx.net> wrote:
Aaron Griffin wrote:
On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> --- makechrootpkg | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/makechrootpkg b/makechrootpkg index 5095425..d1dcf32 100755 --- a/makechrootpkg +++ b/makechrootpkg @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then rm -rf "$uniondir/build/"* fi
+eval $(grep '^SRCDEST=' /etc/makepkg.conf) +eval $(grep '^PKGDEST=' /etc/makepkg.conf) + [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then echo "Setting PKGDEST in makepkg.conf"
The eval seems slightly dangerous to me... does anyone else have this concern, or am I being too careful?
I agree. It is more than "slightly" dangerous. If makepkg.conf contained a line such as: PKGDEST="blabla" && rm -rf / then the eval would indeed execute "rm -rf /". This is particularly bad since makechrootpkg is called with sudo ...
Better look for a safer alternative. What about this?
PKGDEST=$(source /etc/makepkg.conf && echo $PKGDEST) SRCDEST=$(source /etc/makepkg.conf && echo $SRCDEST)
This is not very pretty as we need to source makepkg.conf twice, but at least it is safer. We could also do something like: ORIGDESTDIRS=$(source /etc/makepkg.conf && echo $PKGDEST $SRCDEST) PKGDEST=$(echo $ORIGDESTDIRS | cut -d' ' -f1) SRCDEST=$(echo $ORIGDESTDIRS | cut -d' ' -f2)
I was thinking more along the lines of: Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf) SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
On Thu, Nov 5, 2009 at 13:03, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
I was thinking more along the lines of:
Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf)
SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
I believe that should work... Make sure to throw in a -d -w check after, to make sure it's right... I think a malicious line (".. && rm -rf /") would simply get stored as a string.. as long as we quote everything properly and unset it if it's not right, I'm pretty sure it's not dangerous. It could however, fail in cases where it's split over multiple lines. I don't think that's a reason not to do it though, I can't imagine a reason for such a thing.
On Thu, Nov 5, 2009 at 11:07 AM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
On Thu, Nov 5, 2009 at 13:03, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
I was thinking more along the lines of:
Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf)
SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
I believe that should work... Make sure to throw in a -d -w check after, to make sure it's right... I think a malicious line (".. && rm -rf /") would simply get stored as a string.. as long as we quote everything properly and unset it if it's not right, I'm pretty sure it's not dangerous. It could however, fail in cases where it's split over multiple lines. I don't think that's a reason not to do it though, I can't imagine a reason for such a thing.
Splitting over multiple lines in bash is going to result in a string with embedded \n's in it, which makes an awfully shitty directory name... $ mkdir "foo
bar" $ ls foo bar/ $ rmdir 'foo^Jbar'/
Aaron Griffin wrote:
Aaron Griffin wrote:
On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> --- makechrootpkg | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/makechrootpkg b/makechrootpkg index 5095425..d1dcf32 100755 --- a/makechrootpkg +++ b/makechrootpkg @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then rm -rf "$uniondir/build/"* fi
+eval $(grep '^SRCDEST=' /etc/makepkg.conf) +eval $(grep '^PKGDEST=' /etc/makepkg.conf) + [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then echo "Setting PKGDEST in makepkg.conf"
<snip>
I was thinking more along the lines of:
Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf)
SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
Re: Eric's request of November 11 in another thread, I am in favour of committing the above fix to devtools, and then release a new version. Is it ok with you Aaron? F
On Fri, Nov 20, 2009 at 2:50 AM, Firmicus <Firmicus@gmx.net> wrote:
Aaron Griffin wrote:
Aaron Griffin wrote:
On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> --- makechrootpkg | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/makechrootpkg b/makechrootpkg index 5095425..d1dcf32 100755 --- a/makechrootpkg +++ b/makechrootpkg @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then rm -rf "$uniondir/build/"* fi
+eval $(grep '^SRCDEST=' /etc/makepkg.conf) +eval $(grep '^PKGDEST=' /etc/makepkg.conf) + [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then echo "Setting PKGDEST in makepkg.conf"
<snip>
I was thinking more along the lines of:
Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf)
SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
Re: Eric's request of November 11 in another thread, I am in favour of committing the above fix to devtools, and then release a new version. Is it ok with you Aaron?
F
Bump. Any resolution yet for this SRCDEST/PKGDEST "sourcing" issue?
On Wed, Dec 16, 2009 at 12:17 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
On Fri, Nov 20, 2009 at 2:50 AM, Firmicus <Firmicus@gmx.net> wrote:
Aaron Griffin wrote:
Aaron Griffin wrote:
On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> --- makechrootpkg | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/makechrootpkg b/makechrootpkg index 5095425..d1dcf32 100755 --- a/makechrootpkg +++ b/makechrootpkg @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then rm -rf "$uniondir/build/"* fi
+eval $(grep '^SRCDEST=' /etc/makepkg.conf) +eval $(grep '^PKGDEST=' /etc/makepkg.conf) + [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then echo "Setting PKGDEST in makepkg.conf"
<snip>
I was thinking more along the lines of:
Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf)
SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
Re: Eric's request of November 11 in another thread, I am in favour of committing the above fix to devtools, and then release a new version. Is it ok with you Aaron?
F
Bump. Any resolution yet for this SRCDEST/PKGDEST "sourcing" issue?
Sorry, this has been marked in my little patch queue for too long. I pushed the latest change up there. Let me know if that looks ok to everyone
On Wed, Dec 16, 2009 at 1:56 PM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Wed, Dec 16, 2009 at 12:17 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
On Fri, Nov 20, 2009 at 2:50 AM, Firmicus <Firmicus@gmx.net> wrote:
Aaron Griffin wrote:
Aaron Griffin wrote:
On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
> Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> > --- > makechrootpkg | 3 +++ > 1 files changed, 3 insertions(+), 0 deletions(-) > > diff --git a/makechrootpkg b/makechrootpkg > index 5095425..d1dcf32 100755 > --- a/makechrootpkg > +++ b/makechrootpkg > @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then > rm -rf "$uniondir/build/"* > fi > > +eval $(grep '^SRCDEST=' /etc/makepkg.conf) > +eval $(grep '^PKGDEST=' /etc/makepkg.conf) > + > [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" > if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then > echo "Setting PKGDEST in makepkg.conf" > > <snip>
I was thinking more along the lines of:
Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf)
SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
Re: Eric's request of November 11 in another thread, I am in favour of committing the above fix to devtools, and then release a new version. Is it ok with you Aaron?
F
Bump. Any resolution yet for this SRCDEST/PKGDEST "sourcing" issue?
Sorry, this has been marked in my little patch queue for too long. I pushed the latest change up there.
Let me know if that looks ok to everyone
Thanks. Looks good to me. BTW, do you have this other patch in your queue? http://mailman.archlinux.org/pipermail/arch-dev-public/2009-November/014498....
On Wed, Dec 16, 2009 at 2:05 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
On Wed, Dec 16, 2009 at 1:56 PM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Wed, Dec 16, 2009 at 12:17 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
On Fri, Nov 20, 2009 at 2:50 AM, Firmicus <Firmicus@gmx.net> wrote:
Aaron Griffin wrote:
Aaron Griffin wrote:
> On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote: > > >> Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> >> --- >> makechrootpkg | 3 +++ >> 1 files changed, 3 insertions(+), 0 deletions(-) >> >> diff --git a/makechrootpkg b/makechrootpkg >> index 5095425..d1dcf32 100755 >> --- a/makechrootpkg >> +++ b/makechrootpkg >> @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then >> rm -rf "$uniondir/build/"* >> fi >> >> +eval $(grep '^SRCDEST=' /etc/makepkg.conf) >> +eval $(grep '^PKGDEST=' /etc/makepkg.conf) >> + >> [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" >> if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then >> echo "Setting PKGDEST in makepkg.conf" >> >> <snip>
I was thinking more along the lines of:
Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf)
SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
Re: Eric's request of November 11 in another thread, I am in favour of committing the above fix to devtools, and then release a new version. Is it ok with you Aaron?
F
Bump. Any resolution yet for this SRCDEST/PKGDEST "sourcing" issue?
Sorry, this has been marked in my little patch queue for too long. I pushed the latest change up there.
Let me know if that looks ok to everyone
Thanks. Looks good to me.
BTW, do you have this other patch in your queue? http://mailman.archlinux.org/pipermail/arch-dev-public/2009-November/014498....
Yup, just merged that one too. Thanks
On Wed, Dec 16, 2009 at 1:56 PM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Wed, Dec 16, 2009 at 12:17 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
On Fri, Nov 20, 2009 at 2:50 AM, Firmicus <Firmicus@gmx.net> wrote:
Aaron Griffin wrote:
Aaron Griffin wrote:
On Wed, Nov 4, 2009 at 8:02 PM, Eric Bélanger <snowmaniscool@gmail.com> wrote:
> Signed-off-by: Eric Bélanger <snowmaniscool@gmail.com> > --- > makechrootpkg | 3 +++ > 1 files changed, 3 insertions(+), 0 deletions(-) > > diff --git a/makechrootpkg b/makechrootpkg > index 5095425..d1dcf32 100755 > --- a/makechrootpkg > +++ b/makechrootpkg > @@ -150,6 +150,9 @@ if [ "$REPACK" != "1" ]; then > rm -rf "$uniondir/build/"* > fi > > +eval $(grep '^SRCDEST=' /etc/makepkg.conf) > +eval $(grep '^PKGDEST=' /etc/makepkg.conf) > + > [ -d "$uniondir/pkgdest" ] || mkdir "$uniondir/pkgdest" > if ! grep "PKGDEST=/pkgdest" "$uniondir/etc/makepkg.conf" >/dev/null 2>&1; then > echo "Setting PKGDEST in makepkg.conf" > > <snip>
I was thinking more along the lines of:
Original: eval $(grep '^SRCDEST=' /etc/makepkg.conf)
SRCDEST=$(grep '^SRCDEST=' /etc/makepkg.conf | cut -d= -f2) PKGDEST=$(grep '^PKGDEST=' /etc/makepkg.conf | cut -d= -f2)
Re: Eric's request of November 11 in another thread, I am in favour of committing the above fix to devtools, and then release a new version. Is it ok with you Aaron?
F
Bump. Any resolution yet for this SRCDEST/PKGDEST "sourcing" issue?
Sorry, this has been marked in my little patch queue for too long. I pushed the latest change up there.
Let me know if that looks ok to everyone
You should remove the: # Source global makepkg.conf for SRCDEST and PKGDEST vars [ -f /etc/makepkg.conf ] && source /etc/makepkg.conf [ -f ~/.makepkg.conf ] && source ~/.makepkg.conf near the end of makechrootpkg. It's no longer needed.
Allan McRae schrieb:
Hi,
A new devtools release has been pushed to [testing]. Everything works on a brief test here, but I thought it a good idea to do some testing before pushing to [extra].
Allan
OMGWTFBBQ, this uploads both architectures in one go, I love it. Haven't tested the commit behaviour yet, I got used to running svn commit manually in trunk before running testing/extrapkg due to the bugs in the previous release. However, *pkg seems to work great so far.
Allan McRae wrote:
Hi,
A new devtools release has been pushed to [testing]. Everything works on a brief test here, but I thought it a good idea to do some testing before pushing to [extra].
Is everyone happy with me moving this? We do have the issue of SRCDEST not working that great in makechrootpkg and that can be worked around (was there a patch fixing this?). Other than that, all seems good to me. Allan
On Tue, Nov 10, 2009 at 8:58 PM, Allan McRae <allan@archlinux.org> wrote:
Allan McRae wrote:
Hi,
A new devtools release has been pushed to [testing]. Everything works on a brief test here, but I thought it a good idea to do some testing before pushing to [extra].
Is everyone happy with me moving this? We do have the issue of SRCDEST not working that great in makechrootpkg and that can be worked around (was there a patch fixing this?). Other than that, all seems good to me.
Allan
I did submitted a patch but there was concerns that it might not be safe using eval. Several other ways to do that were proposed but no decision has been made yet. The SRCDEST issue is an annoyance and I have patched my makchrootpkg. So, no objections from me if you want to move it in extra. Eric
participants (7)
-
Aaron Griffin
-
Allan McRae
-
Daenyth Blank
-
Dan McGee
-
Eric Bélanger
-
Firmicus
-
Thomas Bächler