Wed Apr 27 18:39:46 ART 2005 Alberto Bertogli tagged 0.11 Wed Apr 27 18:10:46 ART 2005 Alberto Bertogli * Force the library to be compiled with LFS support. There is a big problem with libraries and LFS support, which happens if you compile the library with LFS support and the application without, or the opposite. In that case the library ABI differs and horrible bugs will happen. This patch prevents it by compiling the library with LFS support always enabled, and put a check in the header to prevent applications using it incorrectly (without LFS). It also updates the manpage to reflect this. Wed Apr 27 18:10:15 ART 2005 Alberto Bertogli * Update the README to add the new manpage. Wed Apr 27 18:04:24 ART 2005 Alberto Bertogli * Add an INSTALL document. Wed Apr 27 18:04:09 ART 2005 Alberto Bertogli * Add manpage support in the Makefile. Wed Apr 27 17:54:06 ART 2005 Alberto Bertogli * Move the license to the doc directory. Wed Apr 27 17:52:03 ART 2005 Alberto Bertogli * Add another stupid test case. Wed Apr 27 17:51:47 ART 2005 Alberto Bertogli * Add a manpage. Wed Apr 27 17:51:07 ART 2005 Alberto Bertogli * Make length match fcntl() semantics. Sun Apr 24 15:52:04 ART 2005 Alberto Bertogli tagged 0.10 Sat Apr 23 22:04:53 ART 2005 Alberto Bertogli * Write something in the README file. Sat Apr 23 21:38:08 ART 2005 Alberto Bertogli * More dirty tests. Sat Apr 23 21:37:03 ART 2005 Alberto Bertogli * Fix filo_plockf() when unlocking. Sat Apr 23 21:33:17 ART 2005 Alberto Bertogli * Put -O3 back into Makefile. Sat Apr 23 21:05:37 ART 2005 Alberto Bertogli * Avoid useless return in plockf(). Sat Apr 23 20:35:56 ART 2005 Alberto Bertogli * Allow sem_wait() to fail, as specified by POSIX. sem_wait() can fail with EINTR, which is easily triggable by stracing the process. The problem is that as of today, Linux manfiles don't list EINTR as a possible error for sem_wait(), so it was a bit tricky to find. I should have learned not to trust manpages by now. Anyway, this is the fix. Sat Apr 23 20:15:28 ART 2005 Alberto Bertogli * Allocate waiter ranges dynamically. This patch makes waiter ranges to be allocated dynamically instead of them living in the thread's stack. See the big comment for details. Sat Apr 23 17:59:29 ART 2005 Alberto Bertogli * Make waiters list FIFO by making it semi-circular. In order to make waiting more fail, make the waiter queue a FIFO, and handle it as a semi-circular list, adding at the end. Sat Apr 23 16:51:36 ART 2005 Alberto Bertogli * Add some dirty tests. Fri Apr 15 19:50:15 ART 2005 Alberto Bertogli * Remove python from the Makefile. Fri Apr 15 19:49:43 ART 2005 Alberto Bertogli * Implement filo_plockf(). Fri Apr 15 14:14:06 ART 2005 Alberto Bertogli * Remove filo_islocked(). Fri Apr 15 02:22:25 ART 2005 Alberto Bertogli * Initial import.