ifile website -> http://www.cs.cmu.edu/~jr6b/ifile/ Thu Apr 22 22:34:54 EDT 1999 ifile-0.7.4.tar.gz I've been unsuccessful in trying to get EXMH to incorporate mail via ifile. I have also been told that the redefinition of EXMH incorporation code has caused incompatibilities, so I've removed such code from ifile.tcl. The INSTALL file has been updated accordingly. If you want to have incorporated mail immedately filtered by ifile, you will need to use slocal or some other filtering tool. Of course, it is still possible to filter mail by binding a key to Msg_Filter or creating a button that calls the Msg_Filter procedure. Tue Feb 9 14:29:28 EST 1999 ifile-0.7.3.tar.gz This release includes a few maintenance fixes: - By default, messages with long attachments are now cropped to reduce processing time. - X-filter: header is added at bottom of header section so that messages in maildrop format are not aversely affected Mon Apr 6 17:06:42 EDT 1998 ifile-0.7.1.tar.gz 0.7.0 included semaphore code which allows only one ifile program per machine to read/write the database at any one point in time. This version includes code which has ifile use a temporary file name while writing the database, renaming this to .idata when the writing has completed. These two fixes should significantly reduce (although not eliminate) problems stemming from multiple ifile jobs being run at the same time. When ifilter.mh passes an e-mail message to the ifile executable, it does so by creating a file in /tmp. Code is included in this version which makes use of the perl unlink() call to remove the file and which makes this file readable only by the user (0600 permissions). Previously, a 'system "rm $file"' call was used to remove the file in /tmp and normal (0644 or 0664) permissions were used for the file. Also included in this release is a minor change which allows one to compile ifile for multiple platforms from the same source (see the --srcdir option of configure). Thu Feb 19 00:23:43 EST 1998 ifile-0.6.6.tar.gz Compilation tested on: Sparc Ultra (SunOS 5.5.1 Generic_103640-12 sun4u sparc SUNW,Ultra-1) DEC (Compaq) Alpha (OSF1 V3.2 148 alpha) Intel Linux (Linux 2.0.27 #5 Mon Jun 2 22:37:38 EDT 1997 i586) Sparc 5 (SunOS 4.1.3_U1 2 sun4m) HP-UX (A.09.05 A 9000/777) Includes a command-line option for turning on/off log files, includes platform specific bugfixes for AIX 4.3 and SunOS 4.1.3 and eliminates dependence on autoconf and m4 for installation (configure.in has been removed). Thu Feb 12 23:42:00 EST 1998 ifile-0.6.5.tar.gz Includes added documentation, two database-related bugfixes, an added command-line option and a Makefile fix. If you have found your ~/.idata file growing unusually large, it is most likely the result of a bug in ifile v0.6.3. This bug is fixed in v0.6.5. Thu Jan 29 00:28:14 EST 1998 ifile-0.6.2.tar.gz Compilation tested on: Sparc Ultra (SunOS 5.5.1 Generic_103640-12 sun4u sparc SUNW,Ultra-1) DEC (Compaq) Alpha (OSF1 V3.2 148 alpha) Intel Linux (Linux 2.0.27 #5 Mon Jun 2 22:37:38 EDT 1997 i586) HP-UX (HP-UX unix30 A.09.05 A 9000/777) Wed Dec 24 23:18:22 EST 1997 ifile-971224.tar.gz I've done a lot of reworking of the organization of ifile. I'm hoping that everything is a bit cleaner and more flexible. Anyway, read INSTALL for installation instructions. As the previous developmental relese did, this one follows the installation scheme which a number of other GNU-style packages follow. Installation will place all the ifile binaries in /usr/local/bin unless you tell configure otherwise. Executables should include ifile, ifilter.mh, irefile.mh, knowledge_base.mh and news2mail. ifile is C code which is not mail client specific. It basically acts as a database front end. ifilter.mh and irefile.mh are perl scripts which interface with ifile to do MH-specific mail handling tasks. knowledge_base.mh is an MH-specific perl script which accumulates information on one's entire mail corpus. knowledge_base.mh is basically a wrapper for ifile and should be much more efficient than previous versions. news2mail is a perl script which Chris Browne whipped up a while ago to filter a news spool through ifile. One assumption this release makes which previous releases have not made is that your MH binaries are in directories which are part of your PATH. If they are not part of your path, you will need to edit ifilter.mh and irefile.mh to give full paths for the MH executables. If you have mail folders which you do *not* want mail filtered to, create an empty .skip_me file in the directories cooresponding to those folders. In previous versions of ifile, knowledge_base was conscious of .skip_me files, but irefile and ifilter ignored them. Now, irefile.mh and ifilter.mh are conscious of .skip_me files (i.e. entries will never be added for those folders and mail will never be filtered to those folders) ifile now has a wealth of lexing options. You might want to try executing "ifile --help" to get a list of the possibilites. The perl scripts all call ifile with the --skip-header (-h) option. ifile uses the alpha-lexer as its default lexer. Note that if you do change lexing styles, you will need to rebuild your database for best results. It is not a good idea to mix and match lexing styles. Also, if you receive mail in non-English languages, you might consider using the white-lexer. It will not throw out =xx (where x is 0-F) sequences, whereas the alpha-lexer will. To make this change, you will need to edit the three MH-specific perl scripts and add "-w" to the $ifile_args variable. As always, if you have any problems, feel free to send me mail at jr6b+@andrew.cmu.edu. ifile has two mailing lists which you are invited to join if you end up becoming a user of ifile. The two mailing lists are ifile-announce@cs.cmu.edu and ifile-discuss@cs.cmu.edu. The discuss mailing list receives all of the announce mail, so you only need to subscribe to one of the two. To subscribe, just send mail to me with a subject of "subscribe" followed by the name of the list you wish to join. As of 97.12.24, I handle these by hand, so you're welcome to say "hi" and tell me what you think of ifile in the subscription e-mail. Sat Oct 18 02:03:11 EDT 1997 ifile-971018.tar.gz Well, I've just completed an overhaul of the database routines. The hash tables are MUCH faster now. I get about a five-fold performance increase. I've also put everything into one executable file, ifile. Do a "./ifile --help" (notice the GNU-style command line parameters :) to get info about all the things the executable will do for you. Most of the options you see work. I think the only one which isn't implemented is --skip-header. Installation is significantly different than before. I've changed to the GNU-style of configure, make, make install. Doing this will compile the binaries and then copy all the executables to a binary directory. You can then call ifile executables (which at this point include "ifile" and "ifilter.mh" - "irefile.mh" will be included before I make a real release :) Sat Oct 4 03:03:39 EDT 1997 ifile-971004.tar.gz I'm in the process of changing the ifile C code that it is mail program independent. I changed the executable names to this effect and changed the nature of the ifilter (now ifile_query) program. irefile (now ifile_db) still operates in the same fashion. To get the effect of an ifilter program, you'll need to use the ifilter.mh script included in the distribution. The idea is that this script interfaces with the ifile_query program and then does all the mail program specific actions necessary to proper result. I also made a MAJOR :) change in the way ifile lexes (parses) messages. I brought in code from another package which does a nice job of abstracting message lexing. The way it's set up, it's not very difficult to change between different styles of lexing and it's also quite possible to write new lexers. My ultimiate goal is to do some testing with different lexing styles. The current lexing style chunks together runs of alphabetic characters (just like older version of ifile did). The knowlege_base.perl script is specific to this style of lexing. This package includes some new hash table and idata writing code, courtesy of Carl Staelin. The process of writing the idata file to disk has been sped up quite significantly. Btw, you'll probably have to edit the perl and C code to make sure that the proper paths are set before you will be able to use this code. $Id: NOTES,v 1.12 1999/04/23 02:50:03 jr6b Exp $