home *** CD-ROM | disk | FTP | other *** search
-
- Release Notes for Columbia MM version 0.88
-
- For 0.88 we tried to make MM more easily portable, using the same
- method that GNU Emacs does. This involves a scheme where, rather than
- trying to figure out which OS's have which features, we instead
- conditionalize the code on feature-specific defines (such as
- HAVE_FLOCK, HAVE_RENAME). This requires a system-describing file
- which contains all the appropriate #define's. So far, we have
- s-ultrix.h, s-sun40.h, s-bsd43.h, s-hpux.h and s-xinu-bsd43.h. See
- the INSTALL file for how to use these. Please let us know if you have
- problems with these. We would appreciate it if you send us any new OS
- files you write (like the pyramid one...).
-
- One serious warning comes with this version of MM. On our Suns
- (running SunOS 4.0), we had a bit of trouble with locking files. If
- you tell MM to use FCNTL for locking (by defining HAVE_F_SETLK in
- s-sunxx.h), but do not run a lock manager daemon (for example, if you
- are trying to lock a file exported (via NFS) from an Ultrix system,
- for which lock managers are not implemented), your MM process will
- hang (in a socket wait) and there is no way to get rid of it. The
- process will be there until the system goes down. Rather unfriendly.
- For this reason, you may wish to fall back on the flock call, which
- unfortunately does not do correct locking across NFS (but at least
- doesn't hang).
-
- We have found one small bug (while trying to announce MM 0.88), which
- should be fixed soon, but decided to release 0.88 since it's about
- time and since it seems to be an infrequent bug. If you have an alias
- which expands to a filename, and the file contains a groupname and
- list of addresses, MM may core dump. The workaround is to not use the
- alias, just type in the file name.
-
- The other warnings are just ones your compiler might give you when you
- compile CCMD. (We got these on our Suns.) Compiling ccmdmd.c you may
- get the warning "illegal pointer combination" since the signal() call
- is sometimes defined to return void. (This is fixed in MM but not yet
- in CCMD.) When ranlib runs, you may get the warning "ccmd.a (dir.o)
- no symbol table" -- this is nothing to worry about.
-
- And now, the new features in MM 0.88.
-
- - The sys-profile program became sys-prof (for short file name SYSV
- types...).
-
- - Many break mask fixes (allowing different characters in certain
- fields).
-
- - We have all new address parsing code that doesn't loop on invalid
- addresses. Instead, it asks if you want to use them (or not,
- depending on the value of "use_invalid_address", which is a yes/no/ask
- type variable). Note that this is only invoked for parsing of
- addresses within messages (when replying, for example) -- CCMD is
- still used to parse addresses from the command line.
-
- - MM now handles the case where the first address displayed is longer
- than the terminal width.
-
- - Escape completion on filenames now completes directory names by
- appending a "/" and pausing, rather than assuming you want the file
- which IS the directory. The "directory-folders" variable can be used
- to get the old behavior on mail-file name parses, and will be most
- useful when MM handles mail formats (mh) where mail files can be
- directories.
-
- - MM now handles an "FCC:" header, which is where all filenames go.
- That is, if you send mail "to *myfile" MM will automatically move that
- to the FCC: field. This prevents problems with replying to mail that
- was sent to a filename and generally looks neater. There is a new
- "default-fcc-list" variable to match "default-bcc-list" and friends,
- as well as "prompt-for-fcc" and an "fcc" command at the S> prompt.
-
- - New "prompt-rcpt-always" variable to always prompt for fcc, bcc, and
- cc (even if a recipient is specified in the line with "send").
-
- - Fix bug in alias parsing -- don't recognize partial matches on alias
- names. So if you have a user named "joe" and an alias "joeshmoe", "to
- joe" will go to user "joe".
-
- - New "compat" module for file locking and other routines to support
- compatibility.
-
- - Redo file locking code to use fcntl when available.
-
- - Make bugsto be bug-mm. It is the responsibility of the installer to
- set up an appropriate "bug-mm" alias on their system (see sys-prof).
-
- - Add hooks for doing usage statistics (define USAGE when compiling to
- get statistics, written to USAGEFILE as defined in pathnames.h).
-
- - Provide for omitting From/Sender/Date headers from input passed to
- sendmail. Define DONT_EMIT_FROM_HEADERS conditional to suppress them.
- (This allows sendmail to add them as it pleases.)
-
- - Preserve protection of mminit over create-init.
-
- - Messages in mbox or mtxt format kept gaining two hours every time MM
- read them in from the file, just a little confusion over Daylight
- Savings Time...
-
- - Use split_args() to take apart editor and speller variables to pass
- argv to exec().
-
- - Various signal and process handling work.
-
- - If display_outgoing_message is true, the last screenful of whatever
- text exists is shown whenever entering text for message body starts
- (that is, whenever a CCMD "paragraph parse" occurs),
-
- - New files created with new file mode.
-
- - Fix problem with autowrap not getting set right from mmfast file.
-
- - In mailrc, understand '\' as a continuation character at end of line
- and allow quoted strings in alias expansions.
-
- - When writing mbox format put a ">" in front of any lines starting
- with "From " since mbox format is so hoky.
-
- - In mbox format, only accept the Flags: field in the header section
- of the message.
-
- - New hostname variables -- see s-xxx.h files.
-
- - The mkversion program now exits with 0 so make is happy.
-
- - Local pathnames (editor and friends) are now set in pathnames.h
- instead of config.h.
-
- - New prepatch program included which looks for patches in the current
- directory and tries to automatically apply them (this may not work
- yet).
-
- - When a sequence is "type"d, it now goes through more as a group,
- rather than one message at a time (with immediate screen clears after
- short messages :-).)
-
- - The "reply" command now allows including/not-including and
- sender/all in either order.
-
- - Fix bug where sender field would get randomly trashed (mostly on
- HP9000's).
-
- - Add unkeyword message sequence to specify all messages without
- keywords.
-
- - Make % mean the same thing as * (last message) in a message sequence, as
- in MM-20.
-
- - Use a more pipe when showing all variables ("show<CR>"), since there
- are SO many.
-
- - Don't abbreviate "/" (root's home dir).
-
- - sys-prof now writes stuff out to ./mm.conf instead of /tmp/mminit.
-
- - The transform program is now more robust, and handles files from
- that OTHER Vax MM :-).
-
-