Hi, On Fri, Aug 29, 2008 at 06:51:55PM +1000, Craig Small wrote: > On Sun, Aug 24, 2008 at 09:37:49AM +0530, Madhusudan.C.S wrote: > > Fixes the PATH_MAX problems in proc/readproc.c file > > and pwdx.c file. This patch builds procps on Hurd. And > > the tools pgrep, pkill, kill, top, watch, tload also > > work. > > This bug report does not explain anything, specifically: > * What IS the PATH_MAX problem? > * Why isn't PATH_MAX defined, its mentioned in the man pages?
I agree that this information (and a quote of the actual build failure) was missing from the initial bug report, so here goes: The GNU system has no path name limit, hence PATH_MAX is not defined. POSIX merely says that PATH_MAX should be defined to the system limit if such a limit exists, which is not the case here. As a result, procps FTBFS on hurd-i386, see http://buildd.debian-ports.org/fetch.php?&pkg=procps&ver=1%3A3.2.7-8&arch=hurd-i386&stamp=1216635885&file=log&as=raw |cc -c -D_GNU_SOURCE -I proc -I/usr/include/ncurses -fno-common -ffast-math -W | -Wall -Wshadow -Wcast-align -Wredundant-decls -Wbad-function-cast | -Wcast-qual -Wwrite-strings -Waggregate-return -Wstrict-prototypes | -Wmissing-prototypes -g -Wdeclaration-after-statement -Wpadded | -Wstrict-aliasing -fno-inline-functions | -DSHARED=0 proc/readproc.c -o proc/readproc.o In file included from |proc/readproc.c:13: |proc/readproc.h:76: warning: padding struct to align 'start_code' |proc/readproc.c: In function 'stat2proc': |proc/readproc.c:391: warning: use of assignment suppression and length modifier | together in scanf format |proc/readproc.c:391: warning: use of assignment suppression and length modifier | together in scanf format |proc/readproc.c: In function 'get_proc_stats': |proc/readproc.c:1039: error: 'PATH_MAX' undeclared (first use in this function) |proc/readproc.c:1039: error: (Each undeclared identifier is reported only once |proc/readproc.c:1039: error: for each function it appears in.) |proc/readproc.c:1039: warning: unused variable 'path' |make[1]: *** [proc/readproc.o] Error 1 |make[1]: Leaving directory `/build/buildd/procps-3.2.7' The reason Madhusudan filed the patch now is that he implemented a procfs translator for the Hurd for Google Summer of Code, so procps actually works fine now on the Hurd; just fixing the FTBFS bug would have been pointless before because there was no /proc in the first place. > * Where is the other source of these programs that now need > alternatives? They are in the `hurd' package (only on hurd-i386, of course). > I'm not shifting essential stuff that procps has around with an > alternatives system until I understand why it needs to happen. At the > very least I'll be checking to see if the alternative files exist (eg > w.procps) before I run update-alternatives As I understand it, Madhusudan just followed the already existing example of w for the other utilities (kill, ps, uptime, vmstat). In any case, this is post-lenny, of course. We'd just be interested in discussing whether you consider the patch OK to apply after the lenny release, so we can make a modified package available to our users. The patch had been reviewed by the Debian GNU/Hurd porters before submission. thanks, Michael -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
