From 72107d51f93406216ec45dc4a7aeec46036cf659 Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Sun, 25 Jan 2009 19:27:04 +0100 Subject: Document the options in the README --- README | 26 +++++++++++++++++++++++--- 1 file changed, 23 insertions(+), 3 deletions(-) (limited to 'README') diff --git a/README b/README index a4a200a..b38911e 100644 --- a/README +++ b/README @@ -8,9 +8,6 @@ versions of the Linux kernel to speed up tailing files in the follow mode (the listens to special events sent by the kernel through the inotify API to determine whether a file needs to be reread. -Currently inotail is not fully compatible to neither POSIX or GNU tail but might -be in the future. - Requirements ------------ - Linux kernel 2.6.13 or higher with CONFIG_INOTIFY enabled @@ -33,6 +30,29 @@ inotail binary to /usr/ and the manpage to /usr/share/man/ respectively type: $ make prefix=/usr install +Compatibility & options +----------------------- +inotail is fully option compatible with current POSIX and GNU tail, though the +obsolescent options present in previous versions of those are not present (e.g. +inotail +/-). The following options are present in inotail for +compatibility reasons (e.g. to use inotail as a tail replacement in scripts) but +do have no effect on inotail besides emiting a warning: + +-s N, --sleep-interval=N inotail does not need a sleep interval because it's + relying on inotify. + +--max-unchanged-stats=N inotail watches the containing directory of the file + when following by name (--follow=name or -F). Thus it + gets notified when a file gets moved or deleted. + +--pid=PID Watching the writer PID ios not possible, because + inotail blocks on the read() call on the inotify fd. + In the worst case, the dying process wouldn't cause an + inotify event, so inotail would never get back from + the read(). + +These options are neither documented in the manpage nor the in-program help. + License ------- inotail is licensed under the terms of the GNU General Public License version 2 -- cgit v1.2.3-54-g00ecf