0.3.5

NOTYETDONE - Internet format addresses default (numbers aren't
rfc822-conforming!)

NOTYETDONE - Mail to non-nodelist nodes should simply be forwarded to
host, maybe it will be capable to sort it out. This also would allow
easy avoidance of big-index problem on 16-bit architectures, by only
including hosts of far-away nets, or even all the other networks.

NOTYETDONE - Translation of ibm graphic chars should cover everything,
and should really translate the stuff to something which looks like it
in ASCII.

- Does better job cleaning up pc graphic chars from names.

- Possibility to use lock files instead of locks.

- No more separate job sequencer.

- Unlocks packet in fpack. Never seen trouble because of this bug but
maybe...

- Built-in idiot-unproof dialer for systems without dial() or
with dial which doesn't work.

- BSD 4.3 support should be there, though it may be crude and
inefficient (so what, all this code is!)  I have no possibility to use
4.2 so you have to tell me if it doesn't work. See NOTES about
changes.

- Added support for user.name@p0.f1.n504.z2.fidonet format, though it
is not yet default. 

- Portability enchancements. Hooks for C-alloca, ndir, some others.

- Lots of code cleanup.

- Seenby will be added automatically, sorted and in correct format.

- rfconfig reads stuff from config file, mostly for scripts. Quite simple
now, but is enough for most purposes.

- Name of bad messages changed from badnn to bad.nn.

- If link fails, file will just be deleted, with error message put
in log file. Before it could leave the packet there, and then unpack
it again next time.

- Options for disabling saving of unpacked stuff, bad packets and messages and
so on.

- Most instructions were moved from INSTALL to config.h and config file.

- Config file now keeps most parameters.

- Creates directories itself if it cannot chdir to it and errno == ENOENT.

- Invoke_rfmail uses parsefnetaddress now.

- sig_t for signal type definition.

- #define HAVE_DIRENT and HAVE_NDIR do something.

0.3.4

- Warning in log about misrouted packets

- Made some fixes to make it pass old lints.

- Merged diffs from jam@hutcs (jari@krkct, Jari Malinen). There were fixes
for rmail and several other modules.

- Point information should be preserved in node_entry now.

- Reports size in telink block. Gives some idea in advance about your
draining disk space ;-)

- Header Fidonet-Flags: private | crash | file-attach |
return-receipt-request | is-return-receipt | audit-request. Audits and
receipts are not supported any way, private is noticed (see fnews.cf:
trash-private, accept-private, command).

- ^AEID: kludge is used to create the message-id if it is found in the message,
this might cut some dupes. Old format is still used if no EID is found.

- ^AINTL is inserted if zones differ, and ^ATOPT and ^AFMPT if either
to or from point is non-zero. I have no idea if they are in correct
format and work at all. TOPT and FMPT override point information found in
INTL, though points are not specified in INTL to my knowledge but saw such
spec some time.

- Bad address messages will be output only if debugging.

- Names are stripped of trailing blanks, didn't happen always before.

- Some messages appear at higher verbosity level than before.

- Some bugs missed my eye in lint output, which might explain some
strange problems appearing sometimes.

0.3.3

- WAITCLEAR. 

- PREWAIT, WAITCLEAR, MAX_SEND_RETRIES. Makes it more configureable, though
configuration file would be far better for this kind of purpose.

- Fixed bugs in sealink, which caused transfers on bad lines
to stop with too many retries message.

- Fixed bug in funpack which could cause terminating too early. It was
a bug in get_string, which didn't read the terminating null char if
string was too long just equal to maximum length (someone had 35 char
name :-), and it read one null more, thought it had no subject, sent
subject as message body to mail and got something else than message magic (2),
and thought here is the end. Also now reports if it got something else
than 2, and stops then.

0.3.2

- Receiver is now Sealink capable (if that means just adding blocknum
and its complement after naks and acks?). It seems to work?  I did it
because I had problems when communicating with Binkley, which says
'timeout' in long transfers, with no apparent reason, message appears
without any delay which could be the reason for such message?  Maybe
this works better?  Sender is not capable of Sealink, though it
wouldn't be difficult?

- Some minor changes.

0.3.1

- Line wrap now according to FSC-0001.TXT.

- rnews can be replaced by command in fnews.cf. This makes possible
forwarding sysop-echos and such restricted echos to selected users or
converting echos to mailing lists.

- 'of (node-specification)' will be now removed from name if it is
there and node-specification specified an existing node. This node-spec
will be saved and used if originator cannot be found for some reason
for reply-to path. Order to use is nodelist, orig in name, origin line
in message, echo feed.

- Zone support has gone giant step forward. For points I need more info,
ie where they are located in fidonet structures. There shouldn't be
net-node combinations anywhere, its all Node structures.

- Packet names have been changed to a very strange format, and files
are stored in in and out directories. Specially the name format may change
in future again, now its really strange.

- Special characters will be removed user names and scandinavian chars will
be converted to a and o. Neither conversion happens in usenet format
messages, but I'll work on it (news software may take care if it itself?).

- REMOVE_ID_IN_PARENTHESIS and SEARCH_NAME_IN_NODELIST are no more
options, they will be forced in this version.

- Some bug fixes.

- Maybe something else too but I dont remember what.

0.2.3

- Echomail origin, seenby and so on will be partially printed by
program, makes maintaining easier as admin doesn't need to change
node numbers in many places, and version number will be up to date.

- Some bugs fixed.
