Re: [Fink-devel] Patch for 10.5 Binary Installer
Evan Broder wrote:
Hi,
I think I've got a patch that fixes the postflight script on 10.5.
There are two things that I fix: first, ps axww seems to for some reason
list lots of processes on 10.5. I will freely admit that I don't
understand the options to ps on OS X or any other platform - I just kept
pulling arguments one at a time until it seemed to do the right thing.
Good catch. To me this looks like a bug in Leopard's /bin/ps: When the
-a flag is present, it ignores the -p pid flag. I have immediately
filed a bug with Apple's bugreporter.
The other change causes pathsetup.sh to run with the privileges,
environment, and shell scripts of the user that is being sudo'd to.
Currently, pathsetup.sh will try to write to /var/root/.profile as the
normal user, which fails for obvious reasons.
The environment variables (in particular $HOME and $USER) with which the
installer runs the postflight script keep changing between OS releases,
sometimes between minor releases. There was a time when the current
version of postflight did the right thing, but on Leopard your
suggestion is looks correct, and it looks correct for Tiger, too.
However, I think this needs some more testing: When I try to execute
this by hand from a root command line prompt, I get weird error messages
(perhaps due to my using tcsh as login shell):
root# sudo -i -u costabel /sw/bin/pathsetup.sh
Badly placed ()'s.
With another command it gets weirder:
root# sudo -i -u costabel /usr/bin/printenv HOME
\312\376\272\276^B^G^C^PEp^L^R: Command not found.
^QU\211\345WVS\203\354,\213u^H\213]^L\353^E\350\261\377\377\377\307D:
Event not found.
There is definitely something fishy about this -i flag when used with
a command. On the other hand, -H seems to work correctly:
root# sudo -H -u costabel /usr/bin/printenv HOME
/Users/costabel
root# sudo -H -u costabel /sw/bin/pathsetup.sh
Setting up your Fink environment
Your environment seems to be correctly
set up for Fink already.
- Evan Broder
cvs diff: Diffing installer/resources
Index: installer/resources/postflight
===
RCS file: /cvsroot/fink/scripts/installer/resources/postflight,v
retrieving revision 1.13
diff -u -r1.13 postflight
--- installer/resources/postflight9 Jun 2006 16:59:14 -1.13
+++ installer/resources/postflight4 Feb 2008 02:32:33 -
@@ -16,7 +16,7 @@
# get ps data for a given PID
sub process_info {
my($pid) = shift;
-my $command = ps axww -p$pid -o user,ppid,tty,command | tail
-n 1;
+my $command = ps xww -p$pid -o user,ppid,tty,command | tail -n 1;
my @info = split(/\s+/, `$command`, 4);
return (
user = $info[0],
@@ -40,7 +40,7 @@
# Running from Installer.app: have Aqua session, so run pathsetup
# as user who ran Installer.app.
my $real_user = $info{user};
-system sudo -u $real_user /sw/bin/pathsetup.sh;
+system sudo -i -u $real_user /sw/bin/pathsetup.sh;
system osascript -e 'tell application \Installer\' -e 'activate'
-e 'end tell';
} else {
# Running from commandline: not certain to have Aqua session, so
--
Martin
-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/
___
Fink-devel mailing list
Fink-devel@lists.sourceforge.net
http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] Patch for 10.5 Binary Installer
Martin Costabel wrote: Good catch. To me this looks like a bug in Leopard's /bin/ps: When the -a flag is present, it ignores the -p pid flag. I have immediately filed a bug with Apple's bugreporter. The environment variables (in particular $HOME and $USER) with which the installer runs the postflight script keep changing between OS releases, sometimes between minor releases. There was a time when the current version of postflight did the right thing, but on Leopard your suggestion is looks correct, and it looks correct for Tiger, too. However, I think this needs some more testing: When I try to execute this by hand from a root command line prompt, I get weird error messages (perhaps due to my using tcsh as login shell): root# sudo -i -u costabel /sw/bin/pathsetup.sh Badly placed ()'s. With another command it gets weirder: root# sudo -i -u costabel /usr/bin/printenv HOME \312\376\272\276^B^G^C^PEp^L^R: Command not found. ^QU\211\345WVS\203\354,\213u^H\213]^L\353^E\350\261\377\377\377\307D: Event not found. There is definitely something fishy about this -i flag when used with a command. On the other hand, -H seems to work correctly: root# sudo -H -u costabel /usr/bin/printenv HOME /Users/costabel root# sudo -H -u costabel /sw/bin/pathsetup.sh Setting up your Fink environment Your environment seems to be correctly set up for Fink already. Yeah, -H is probably what you want more than -i. In terms of behavior changing across versions, this isn't something we should run into again - by using -H you explicitly specify the behavior you want; this is a flag that is standardized across every sudo I've run into. - Evan - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] Patch for 10.5 Binary Installer
Martin Costabel wrote: Good catch. To me this looks like a bug in Leopard's /bin/ps: When the -a flag is present, it ignores the -p pid flag. I have immediately filed a bug with Apple's bugreporter. I imagine that it will be closed as invalid. With -a, you are essentially asking for all processes with a controlling terminal, with -p you ask for all processes which match the given process id. ps will return both all the processes with a controlling terminal and all processes matching the pid. $ ps -a | wc -l 4 $ ps -a -p 1 | wc -l 5 http://www.opengroup.org/onlinepubs/95399/utilities/ps.html Peter -- Peter O'Gorman http://pogma.com - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
[Fink-devel] Fwd: Fink autossh package
I am fowarding this to fink-devel because maintainer's email bounces. Sébastien Begin forwarded message: From: Sébastien Maret [EMAIL PROTECTED] Date: February 4, 2008 16:17:31 CEST To: [EMAIL PROTECTED] Subject: Fink autossh package Hi, Fink autossh package is a bit outdated. Would you mind updating it (or do you mind if I update it for you)? Thanks, Sébastien - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] Fwd: Fink autossh package
Sébastien Maret wrote: I am fowarding this to fink-devel because maintainer's email bounces. Sébastien Begin forwarded message: From: Sébastien Maret [EMAIL PROTECTED] Date: February 4, 2008 16:17:31 CEST To: [EMAIL PROTECTED] Subject: Fink autossh package Hi, Fink autossh package is a bit outdated. Would you mind updating it (or do you mind if I update it for you)? Thanks, Sébastien I guess you might as well do the update and commit it in a few days if the maintainer hasn't replied. - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
Re: [Fink-devel] Patch for 10.5 Binary Installer
Peter O'Gorman wrote: [] I imagine that it will be closed as invalid. With -a, you are essentially asking for all processes with a controlling terminal, with -p you ask for all processes which match the given process id. ps will return both all the processes with a controlling terminal and all processes matching the pid. Argh; I really should know better by now than to read man pages on a Leopard machine on which I haven't mucked out personally the 4500 or so man pages that remain from Tiger when doing an upgrade install. :-( I guess the line proposed by Evan - using ps xww -p pid is correct for Tiger and Leopard anyway. -- Martin - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
[Fink-devel] Fwd: Trying to port your package to Fink
Begin forwarded message: From: Murali Vadivelu [EMAIL PROTECTED] Date: 5 February 2008 02:10:23 GMT To: pguyot AT kallisys.net Subject: Trying to port your package to Fink Dear Developer, I am trying to port you package to Fink, i386-elf-gcc. I am not sure how t troubleshoot this error: ed -e 's/\(UNUSED\)/_GLIBCXX_\1/g' \ -e 's/\(GCC[ABCDEFGHIJKLMNOPQRSTUVWXYZ_]*_H\)/_GLIBCXX_\1/g' \ -e 's/SUPPORTS_WEAK/__GXX_WEAK__/g' \ -e 's/\([ABCDEFGHIJKLMNOPQRSTUVWXYZ_]*USE_WEAK\)/_GLIBCXX_\1/g' \ -e 's,^#include \(.*\),#include bits/\1,g' \ /sw/src/fink.build/i386-elf-gcc-3.4.6-1/gcc-3.4.6/libstdc++- v3/../gcc/gthr-single.h i386-elf/bits/gthr-default.h /bin/sh ./../gcc-3.4.6/mkinstalldirs i386-elf/libiberty ; \ rm -f i386-elf/libiberty/Makefile || : ; \ cp multilib.out i386-elf/libiberty/multilib.out mkdir i386-elf/libiberty Configuring in i386-elf/libiberty configure: creating cache ./config.cache checking whether to enable maintainer-specific portions of Makefiles... no checking for makeinfo... makeinfo checking for perl... perl checking build system type... i686-apple-darwin9.1.0 checking host system type... i386-pc-elf checking for i386-elf-ar... i386-elf-ar checking for i386-elf-ranlib... i386-elf-ranlib checking for i386-elf-gcc... /sw/src/fink.build/i386-elf- gcc-3.4.6-1/build/gcc/xgcc -B/sw/src/fink.build/i386-elf-gcc-3.4.6-1/ build/gcc/ -B/sw/share/i386-elf/i386-elf/bin/ -B/sw/share/i386-elf/ i386-elf/lib/ -isystem /sw/share/i386-elf/i386-elf/include -isystem / sw/share/i386-elf/i386-elf/sys-include checking for suffix of object files... o checking whether we are using the GNU C compiler... yes checking whether /sw/src/fink.build/i386-elf-gcc-3.4.6-1/build/gcc/ xgcc -B/sw/src/fink.build/i386-elf-gcc-3.4.6-1/build/gcc/ -B/sw/ share/i386-elf/i386-elf/bin/ -B/sw/share/i386-elf/i386-elf/lib/ - isystem /sw/share/i386-elf/i386-elf/include -isystem /sw/share/i386- elf/i386-elf/sys-include accepts -g... yes checking for /sw/src/fink.build/i386-elf-gcc-3.4.6-1/build/gcc/xgcc -B/sw/src/fink.build/i386-elf-gcc-3.4.6-1/build/gcc/ -B/sw/share/ i386-elf/i386-elf/bin/ -B/sw/share/i386-elf/i386-elf/lib/ -isystem / sw/share/i386-elf/i386-elf/include -isystem /sw/share/i386-elf/i386- elf/sys-include option to accept ANSI C... none needed checking how to run the C preprocessor... /sw/src/fink.build/i386- elf-gcc-3.4.6-1/build/gcc/xgcc -B/sw/src/fink.build/i386-elf- gcc-3.4.6-1/build/gcc/ -B/sw/share/i386-elf/i386-elf/bin/ -B/sw/ share/i386-elf/i386-elf/lib/ -isystem /sw/share/i386-elf/i386-elf/ include -isystem /sw/share/i386-elf/i386-elf/sys-include -E checking whether /sw/src/fink.build/i386-elf-gcc-3.4.6-1/build/gcc/ xgcc -B/sw/src/fink.build/i386-elf-gcc-3.4.6-1/build/gcc/ -B/sw/ share/i386-elf/i386-elf/bin/ -B/sw/share/i386-elf/i386-elf/lib/ - isystem /sw/share/i386-elf/i386-elf/include -isystem /sw/share/i386- elf/i386-elf/sys-include and cc understand -c and -o together... yes checking for an ANSI C-conforming const... yes checking for inline... inline checking whether byte ordering is bigendian... cross-compiling... unknown checking to probe for byte ordering... /sw/share/i386-elf/i386-elf/ bin/ld: crt0.o: No such file: No such file or directory collect2: ld returned 1 exit status unknown configure: error: unknown endianess - sorry /sw/src/fink.build/i386-elf-gcc-3.4.6-1/gcc-3.4.6/libiberty/ configure: line 3289: exit: please: numeric argument required /sw/src/fink.build/i386-elf-gcc-3.4.6-1/gcc-3.4.6/libiberty/ configure: line 3289: exit: please: numeric argument required make: *** [configure-target-libiberty] Error 1 ### execution of /var/tmp/tmp.1.BNrDZG failed, exit code 2 Please help if you could. With many thanks. Best, Murali. - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/___ Fink-devel mailing list Fink-devel@lists.sourceforge.net http://news.gmane.org/gmane.os.apple.fink.devel
