Start of man-pages-3.24: updating Changes and Changes.old

This commit is contained in:
Michael Kerrisk 2009-09-30 05:21:56 +02:00
parent dc7e17a319
commit 52725cd997
2 changed files with 205 additions and 169 deletions

187
Changes
View File

@ -1,6 +1,6 @@
==================== Changes in man-pages-3.23 ====================
==================== Changes in man-pages-3.24 ====================
Released: 2009-09-30, Munich
Released: ????-??-??, Munich
Contributors
@ -9,177 +9,26 @@ Contributors
The following people contributed notes, ideas, or patches that have
been incorporated in changes in this release:
Aaron Gardner <aaron.j.gardner@gmail.com>
Andrey Vihrov <vihrov@gmail.com>
Christoph Hellwig <hch@lst.de>
Georg Sauthoff <gsauthof@techfak.uni-bielefeld.de>
Leslie P. Polzer <sky@viridian-project.de>
Marc Lehmann <debian-reportbug@plan9.de>
Mark Hills <mark@pogo.org.uk>
Michael Kerrisk <mtk.manpages@gmail.com>
Mike Frysinger <vapier@gentoo.org>
Nicolas François <nicolas.francois@centraliens.net>
Serge Hallyn <serue@us.ibm.com>
Siward de Groot <siward@ziggo.nl>
rui rlex <rui.rlex@gmail.com>
Apologies if I missed anyone!
New and rewritten pages
-----------------------
Newly documented interfaces in existing pages
---------------------------------------------
New and changed links
---------------------
Global changes
--------------
Changes to individual pages
---------------------------
execve.2
pipe.2
tee.2
fmemopen.3
mq_notify.3
qsort.3
Michael Kerrisk
Replace use of assert() by code that checks argc
See http://bugzilla.kernel.org/show_bug.cgi?id=13569
As noted by Andrey:
The purpose of the assert macro, defined in <assert.h>,
is to provide a tool to check for programming mistakes
or program logic errors. However, the assert macro must
never be used to perform checks for run time errors,
since, with the NDEBUG macro defined, expressions within
the assert macro invocations are not evaluated/checked
for, resulting in behavior that was not originally intended.
...
The pages affected in the core package are
execve(2)
pipe(2)
tee(2)
fmemopen(3)
mq_notify(3)
qsort(3)
getrusage.2
Michael Kerrisk
ru_inblock and ru_oublock are now implemented
These fields of the rusage structure are filled in since
Linux 2.6.22.
mmap.2
Michael Kerrisk
Add brief documentation of MAP_HUGETLB
This flag is new in 2.6.32, and serves a similar
purpose to the shmget() SHM_HUGETLB flag.
open.2
Christoph Hellwig
add some comments on O_SYNC and friends
poll.2
Michael Kerrisk
Clarify wording describing of 'nfds' argument.
reported by: rui rlex <rui.rlex@gmail.com>
semctl.2
Nicolas François
Remove some redundant words
setpgid.2
Michael Kerrisk
Add an explanation of orphaned process groups
splice.2
tee.2
vmsplice.2
Mark Hills
Fix return type
Since glibc 2.7, the return type for these functions
is ssize_t (formerly it was long).
stat.2
Nicolas François
Fix small bug in example program
Since it is a failure, EXIT_FAILURE looks more appropriate
than EXIT_SUCCESS.
umount.2
Michael Kerrisk
glibc only exposes MNT_DETACH and MNT_EXPIRE since version 2.11
See http://sourceware.org/bugzilla/show_bug.cgi?id=10092
exit.3
Michael Kerrisk
Add a pointer to explanation of orphaned process groups in setpgid(2)
fflush.3
Michael Kerrisk
fflush() discards buffered input
ffs.3
Michael Kerrisk
Clarify that ffsl() and ffsll() are GNU extensions
getaddrinfo.3
Michael Kerrisk
Note nonstandard assumed hints.ai_flags value when hints is NULL
When hints is NULL, glibc assumes hints.ai_flags is
AI_V4MAPPED|AI_ADDRCONFIG whereas POSIX says 0.
According to Ulrich Drepper, glibc's behavior is better.
getmntent.3
Mike Frysinger
setmntent() argument is 'filename' not 'fp'
The description of setmntent() formerly used the wrong
argument name.
posix_fallocate.3
Nicolas François
Fix reference to POSIX.1-2008
The sentence mentions twice POSIX.1-2001.
I guess the second one should be POSIX.1-2008.
This should be checked in the standard.
setenv.3
Michael Kerrisk
Improve ERRORS section
Add ENOMEM error; improve EINVAL description. Also, make
RETURN VALUE section a little more accurate in its mention
of errno.
strftime.3
Nicolas François
Fix error in description: s/Monday/Thursday/
proc.5
Nicolas François
Fix page cross reference
max_user_watches is better explained in epoll(7) than inotify(7).
proc.5
Michael Kerrisk
dmesg is in section 1, not section 8
capabilities.7
Michael Kerrisk
FS UID manipulations affect CAP_LINUX_IMMUTABLE and CAP_MKNOD
Nowadays, file system UID manipulations also affect
CAP_LINUX_IMMUTABLE (since 2.6.3) and CAP_MKNOD (since 2.6.29).
capabilities.7
Michael Kerrisk
Fix version number for CAP_MKNOD in DS UID manipulations
A recent patch said "since 2.6.29". It should have
been "since 2.6.30".
capabilities.7
Nicolas François
Reword a bad sentence in description of capability bounding set.
mq_overview.7
Michael Kerrisk
Change documented ranges for msg_max and msgsize_max
Linux 2.6.28 changed the permissible ranges for
these /proc files.
tcp.7
udp.7
Nicolas François
Replace references to syctl interfaces with /proc

View File

@ -18844,3 +18844,190 @@ standards.7
Gallmeister and Lewine are rather old books. Probably,
there are better books to consult nowadays, and anyway,
this man page isn't intended to be a bibliography.
==================== Changes in man-pages-3.23 ====================
Released: 2009-09-30, Munich
Contributors
------------
The following people contributed notes, ideas, or patches that have
been incorporated in changes in this release:
Aaron Gardner <aaron.j.gardner@gmail.com>
Andrey Vihrov <vihrov@gmail.com>
Christoph Hellwig <hch@lst.de>
Georg Sauthoff <gsauthof@techfak.uni-bielefeld.de>
Leslie P. Polzer <sky@viridian-project.de>
Marc Lehmann <debian-reportbug@plan9.de>
Mark Hills <mark@pogo.org.uk>
Michael Kerrisk <mtk.manpages@gmail.com>
Mike Frysinger <vapier@gentoo.org>
Nicolas François <nicolas.francois@centraliens.net>
Serge Hallyn <serue@us.ibm.com>
Siward de Groot <siward@ziggo.nl>
rui rlex <rui.rlex@gmail.com>
Apologies if I missed anyone!
Changes to individual pages
---------------------------
execve.2
pipe.2
tee.2
fmemopen.3
mq_notify.3
qsort.3
Michael Kerrisk
Replace use of assert() by code that checks argc
See http://bugzilla.kernel.org/show_bug.cgi?id=13569
As noted by Andrey:
The purpose of the assert macro, defined in <assert.h>,
is to provide a tool to check for programming mistakes
or program logic errors. However, the assert macro must
never be used to perform checks for run time errors,
since, with the NDEBUG macro defined, expressions within
the assert macro invocations are not evaluated/checked
for, resulting in behavior that was not originally intended.
...
The pages affected in the core package are
execve(2)
pipe(2)
tee(2)
fmemopen(3)
mq_notify(3)
qsort(3)
getrusage.2
Michael Kerrisk
ru_inblock and ru_oublock are now implemented
These fields of the rusage structure are filled in since
Linux 2.6.22.
mmap.2
Michael Kerrisk
Add brief documentation of MAP_HUGETLB
This flag is new in 2.6.32, and serves a similar
purpose to the shmget() SHM_HUGETLB flag.
open.2
Christoph Hellwig
add some comments on O_SYNC and friends
poll.2
Michael Kerrisk
Clarify wording describing of 'nfds' argument.
reported by: rui rlex <rui.rlex@gmail.com>
semctl.2
Nicolas François
Remove some redundant words
setpgid.2
Michael Kerrisk
Add an explanation of orphaned process groups
splice.2
tee.2
vmsplice.2
Mark Hills
Fix return type
Since glibc 2.7, the return type for these functions
is ssize_t (formerly it was long).
stat.2
Nicolas François
Fix small bug in example program
Since it is a failure, EXIT_FAILURE looks more appropriate
than EXIT_SUCCESS.
umount.2
Michael Kerrisk
glibc only exposes MNT_DETACH and MNT_EXPIRE since version 2.11
See http://sourceware.org/bugzilla/show_bug.cgi?id=10092
exit.3
Michael Kerrisk
Add a pointer to explanation of orphaned process groups in setpgid(2)
fflush.3
Michael Kerrisk
fflush() discards buffered input
ffs.3
Michael Kerrisk
Clarify that ffsl() and ffsll() are GNU extensions
getaddrinfo.3
Michael Kerrisk
Note nonstandard assumed hints.ai_flags value when hints is NULL
When hints is NULL, glibc assumes hints.ai_flags is
AI_V4MAPPED|AI_ADDRCONFIG whereas POSIX says 0.
According to Ulrich Drepper, glibc's behavior is better.
getmntent.3
Mike Frysinger
setmntent() argument is 'filename' not 'fp'
The description of setmntent() formerly used the wrong
argument name.
posix_fallocate.3
Nicolas François
Fix reference to POSIX.1-2008
The sentence mentions twice POSIX.1-2001.
I guess the second one should be POSIX.1-2008.
This should be checked in the standard.
setenv.3
Michael Kerrisk
Improve ERRORS section
Add ENOMEM error; improve EINVAL description. Also, make
RETURN VALUE section a little more accurate in its mention
of errno.
strftime.3
Nicolas François
Fix error in description: s/Monday/Thursday/
proc.5
Nicolas François
Fix page cross reference
max_user_watches is better explained in epoll(7) than inotify(7).
proc.5
Michael Kerrisk
dmesg is in section 1, not section 8
capabilities.7
Michael Kerrisk
FS UID manipulations affect CAP_LINUX_IMMUTABLE and CAP_MKNOD
Nowadays, file system UID manipulations also affect
CAP_LINUX_IMMUTABLE (since 2.6.3) and CAP_MKNOD (since 2.6.29).
capabilities.7
Michael Kerrisk
Fix version number for CAP_MKNOD in DS UID manipulations
A recent patch said "since 2.6.29". It should have
been "since 2.6.30".
capabilities.7
Nicolas François
Reword a bad sentence in description of capability bounding set.
mq_overview.7
Michael Kerrisk
Change documented ranges for msg_max and msgsize_max
Linux 2.6.28 changed the permissible ranges for
these /proc files.
tcp.7
udp.7
Nicolas François
Replace references to syctl interfaces with /proc