[lsb-discuss] Update 1: The Plan

Steve Schafer sschafer at freestandards.org
Wed Jan 10 10:53:04 PST 2007


Thanks, Stew. 

> -----Original Message-----
> From: Stew Benedict [mailto:sbenedict at mandriva.com] 
> Sent: Wednesday, January 10, 2007 5:59 AM
> To: Steve Schafer
> Cc: lsb-discuss at freestandards.org
> Subject: Re: [lsb-discuss] Update 1: The Plan
> 
> 
> On Tue, 9 Jan 2007, Wichmann, Mats D wrote:
> 
> > 
> > lsb-runtime-test: still being worked on
> 
> Some details on runtime.
> 
> I'm still fighting the perms issue. Whatever mechanism we 
> relied on in the past doesn't seem to be recognized by 
> current rpm and just putting the files in the package without 
> specifying ownership ends up with everything owned by the 
> user that built the package, in this case root.
> 
> We copy "everything" out of the build dir (/home/tet) with 
> rsync to /var/tmp. At this point the permissions look 
> reasonable, so it's something rpm is doing internally at the 
> %files stage where they change.
> 
> I'm now tweaking %files to specify those special files that 
> need root.root or root.vsxg0 ownership. Seems better that we 
> have a documented list than rely on things just being right somehow.
> 
> Also, since we rsync the whole tree, rpm spews out a huge 
> list of "unpackaged files" which just adds to the build 
> noise.  I've gone back now and changed things to just rsync 
> the bits we intend to package.
> 
> Finally, in comparing my rpm to the official 3.1.0-6, I see 
> I'm missing 40-some files, primarily in LSB.pam/testcases, 
> and POSIX.os/procprim/exec, which probably explains the 
> dismal test results I've been getting. I'm also missing some 
> files in test_sets/INC, which don't even seem to be listed in 
> %files in the spec, so I'm not sure how they got in the 
> previous build, or whether they're needed.
> 
> Unfortunately, there is little clue as to why they aren't 
> built when I try to capture the build output.  I have now 
> installed pam-devel, hoping that might help and am doing 
> another run, the only other clues I see in the log is:
> 
> *** Starting pthread.h
> Missing: symbol matching /CLK_TCK/
> Missing: symbol matching /CLOCK_REALTIME/
> Missing: symbol matching /CLOCKS_PER_SEC/
> Missing: symbol matching /NULL/
> Missing: symbol matching /SCHED_FIFO/
> Missing: symbol matching /SCHED_OTHER/
> Missing: symbol matching /SCHED_RR/
> Missing: symbol matching /TIMER_ABSTIME/
> *** Completed pthread.h
> 
> *** Starting signal.h
> Missing: #define        NSIG    (-1)    /* user supplied: 
> (highest_signal_number + 1) */
> *** Completed signal.h
> 
> *** Starting sys/time.h
> Missing: struct itimerval { <members> };
> Missing: struct timeval { <members> };
> *** Completed sys/time.h
> 
> Grepping under /usr/include finds these on my system, so I'm 
> not quite sure what the problem is.
> 
> If anyone with more experience with this beast has 
> suggestions, I'm listening :)
> 
> --
> Stew Benedict
> 
> 




More information about the lsb-discuss mailing list