changes

oscar message protocol stack

 - Thu Mar 29 13:05:50 PST 2001
  - Less leaks.

 - Thu Mar 29 14:45:22 CST 2001 (jbm)
  - Add module capability to the 0001/001f stuff (jbm)

 - Wed Mar 28 21:20:08 PST 2001
  - Add flags arg to aim_chat_send_im()
  - Add msglen arg to aim_send_im and aim_chat_send_im
     - This will make more sense when I make the UNICODE commit (tomorrow?)

 - Wed Mar 28 16:51:25 PST 2001
  - I decided it was a good day while I was figuring out that it was a bad day.
  - See faimtest and login.c::memrequest().
  - Hopefully I'm done for a while.

 - Tue Mar 27 19:15:10 PST 2001
  - I haven't decided yet whether today is a good day or a bad day.
  - Probably both.
  - More later.

 -  Mon Mar 26 16:08:45 PST 2001
  - Why didn't anyone tell me buddy-offgoing wasn't working?
    - *** REQUIRES CLIENT CHANGES... buddy-offgoing now passes a userinfo
          struct (its now identical to buddy-oncoming)

 - Mon Mar 26 15:57:41 PST 2001
  - Hrmph.
  - Debuglevel wasn't getting set properly by aim_session_init

 - Sat Mar 24 03:16:32 UTC 2001
  - vaargs to msgerror and locateerror swapped
     - ** CLIENT CHANGE: Reason code comes before SN now.  This is so that
          the backend code for all three error handlers is identical.
  - aim_getinfo() returns -1 on failure (hah! I don't think anyone checks 
      return codes from libfaim anyway...they're known for their
      inconsistency and lack of sensicality)
  - Remove _DEFAULT handlers in faimtest...
     - They were not consistent.
     - They did not make sense.
     - They did not work anyway.
     - Most of them were #if 0'd.
  - Add ICBM Parameter Info callback... you can catch this if you want.
  - aim_getbuildstring() now takes a buffer instead of returning static (ew!)
  - Clean up, "factorize", etc
  - More stuff... (Only one "middle handler" left in rxhandlers.c!)

 - Fri Mar 23 05:42:11 UTC 2001
  - Export aim_encode_password_md5()
  - Add middle handler for 000b/0002 (min report interval)
  - Add aim_session_kill()
     - CLIENTS MUST CALL THIS either in addition to or instead of
       aim_logoff(), particularly if you keep lots of sessions open.
       (Sessions now contain dynamically allocated memory which must be freed.)
  - Oh, and some other stuff...

 - Fri Mar 23 01:45:28 UTC 2001
  - Add AIM_CLIENTINFO_KNOWNGOOD
  - Fix a few details in aim_send_login (thanks temas)
  - Prevent aim_rxhandlers from being called recursively
     - This is something that has been in my PENDING_AIM mailbox for over
       a year now.  I figured it was trivial enough to throw in.  EveryBuddy
       wanted it, but I don't remember why.  It should never happen anyway.
  - Use AIM_CLIENTINFO_KNOWNGOOD
     - This lets faimtest log in after today's round of AOL being mean.

 - Fri Mar 16 23:12:58 UTC 2001
  - Remove one level from aim_rxdispatch()

 - Wed Mar 14 20:21:49 UTC 2001
  - Clean up declarations of TLV functions (ie, const)
  - Add -o and -O to faimtest (I didn't use your patch, josh)

 - Wed Mar 14 03:11:03 UTC 2001
  - Err...I'll commit before I go home.  More faimtest crap.

 - Tue Mar 13 20:23:04 UTC 2001
  - Fiddle with faimtest a bit.

 - Sun Mar 11 06:02:19 UTC 2001
  - Banish socklen_t.

 - Sun Mar 11 05:51:45 UTC 2001 (jbm)
  - Split off aim_internal.h 
  - Add aim_tx_cleanqueue()
  - Remove sn from directim callback (use priv)

 - Mon Mar  5 01:19:48 UTC 2001
  - Fix typo in last CHANGES entry.
  - Add fix from Eric to keep efence from complaining about malloc(0)
  - Add .cvsignore files (also from Eric)
  - Throw in josh's latest OFT/ODC kludges
  - Completly remove all calls to printf and (unbounded) sprintf
  - Add aim_setdebuggingcb()
  - Remove preprocessor-based debug semantics (faimconfig.h)
  - faimdprintf() now calls a user-specified callback to do printing
  - Rearrange aim_tx_new() parameters to be more sensical, add sess
  - Do some sanity checking in aim_tx_new()
  - Add #defines to keep printf from compiling inside libfaim
  - Who knows what else. (in total, this patch is nearly 5k lines)

 - Fri Mar  2 02:17:22 UTC 2001
  - Add aim_fingerprintclient(). (AB will trust this, but you shouldn't.)
  - Add some EXTRA_DIST to Makefile.am

 - Mon Feb 26 01:46:34 UTC 2001
  - Fix swapped bytes in aim_caps[]. Thanks to 
      Bruce Miller <[e-mail]> for pointing this out.

 - Sun Feb 11 01:07:36 UTC 2001
  - Make secondary connections to the authorizer work again
  - Make aim_auth_changepasswd() work again.
  - Add aim_auth_setversions(), aim_auth_getinfo(), aim_auth_setemail(),
    aim_auth_reqconfirm()
  - Add middle handlers for responses to all of above.
  - Add examples of above to faimtest (start with 'reqauth')
 
 - Fri Feb  9 22:31:22 UTC 2001
  - Add aim_tx_setenqueue().

 - Thu Feb  8 20:12:39 UTC 2001
  - Let TLV functions out of aim.h if FAIM_NEED_TLV is defined.

 - Thu Feb  8 02:31:25 UTC 2001
  - Remove aim_purgecookies() -- never used anyway
  - Clean up a few bits of aim_msgcookie.c
  - Remove AIM_CONN_MAX constant (hasn't been relevent in ages)
  - Change commentary on synchro in aim.h
  - Add NOPLOCKS and make them default
  - Make faim/aim.h more of a mess; try to keep interfaces hiddenish.
  - Add #define FAIM_INTERNAL to top of every file
  - Remove aim_countconn (no one uses it anyway)
  - Add aim_cloneconn() (cringe.)
  - Throw in josh's getfile patch (and then munge it enough to break it)
  - Add getopt() to faimtest.
  - Remove aim_buildcode.h/mkbuildinfo.sh crap.
  - Clean up aim_cleansnacs()
  - There may have been other stuff too.  As always, the diff is the 
    most accurate changelog there is.
  - *** FT is currently more broken than it was before.  Also, this patch
    introduces even more memory leaks.  You've been warned.

 - Sun Dec 17 07:19:04 UTC 2000
  - Update the capability block list

 - Sat Dec 16 01:34:19 UTC 2000
  - Rename paramid to rateclass
  - Add a long comment about the rate system
  - Add a few other comments
  - Clean up aim_bos_clientready().

 - Fri Dec 15 23:35:01 UTC 2000
  - Add genericerr callback/middle. 
   - This is for catching the errors that come back, for instance, from
     sending a SNAC that isn't supported by the host (such as the
     Watcher Request SNAC).

 - Fri Dec 15 21:51:32 UTC 2000
  - Add more info to evilnotify callback
  - Add a few values to aim_cbtypes.h

 - Fri Dec 15 20:41:15 UTC 2000
  - Revise README
  - Make aimdebugd compile
  - Typos

 - Fri Dec 15 02:04:12 UTC 2000
  - Parse rate changes *properly*
  - Add sample code to faimtest demonstrating my approximations
      on how to fiddle with aim_conn_setlatency() in relation
      to the rate changes (properly wildly conservative). [I also
      copied this stuff into mfaim.]

 - Thu Dec 14 03:39:34 UTC 2000
  - Change the 0x01 to a 0x00 in auth request to make buddy lists work again
  - Rearrange bytes in setversions() to match winaim
  - Add several cachesnac() calls
  - Add hostonline and hostversions parsers (superfluous)

 - Wed Dec 13 02:26:39 UTC 2000
  - Create aim_cachesnac()
  - Some long overdue cleanups

 - Wed Dec 13 00:38:56 UTC 2000
  - Added the client update/version info to the authresp callback
  - Added the prototype for aim_getfile_send_chunk to aim.h

 - Tue Dec 12 23:02:41 UTC 2000
  - Got pissed off at sess->logininfo.  Got rid of it.
    - Now pass all that stuff in as varargs, like it should be.
    - *** Look at the changes to faimtest. You'll also need to
          change anything in your code that references sess->logininfo
          to reference sess->sn instead.  The rest of the other info
          is now unavailable (it was before, too, it just didnt look like it).
  - A few other minor cleanups.
  - Added aim_gettlv8/16/32, aim_puttlv_8, and aim_addtlvtochain_noval.
  - Added that short 0x004a TLV to the auth request, like WinAIM 4.3.

 - Mon Dec  4 23:46:35 UTC 2000
  - Add exchange to the create response callback (doh!)

 - Sat Dec  2 02:14:45 UTC 2000
  - Dumb bug in aim_clearhandlers()

 - Fri Dec  1 23:48:38 UTC 2000
  - Store exchange along with room name in the great chat hack

 - Fri Dec  1 22:25:56 UTC 2000
  - Fix numerous tiny (but sometimes catastrophic) bugs dealing
      with connection death (particularly with chat connections)
  - *** Any connection with a -1 fd will get returned by aim_select
      immediatly now... your code probably already handles this implicitly.

 - Wed Nov 29 17:31:23 UTC 2000
  - Rewrote some of the msgcookie stuff
  - Changed cachecookies to uncachecookies where it makes sense (arg!)
  - Minor other stuff
 
 - Fri Nov 10 08:24:34 UTC 2000
  - Add sess->flags (replaces sess->snaclogin)
  - Remove odd setstatus call in chat parser
  - Remove aim_tx_enqueue macro, replace with a smarter one
    - If a connection is in progress, enqueue instead of calling
      the client-specified queuer
  - Add support for nonblocking connects through the 
     AIM_SESS_FLAG_NONBLOCKCONNECT flag to aim_session_init()
  - Add AIM_CB_SPECIAL_CONNCOMPLETE callback. Not real useful.
  - Add AIM_CB_SPECIAL_FLAPVER callback. Can be used as an alternate
     way of starting the login process, or just to look like you 
     know what you're doing. Fixed associated inconsistencies
     in aim_rxhandlers too.
  - Fix some connection status stupidities in faimtest.

 - Wed Nov  8 13:11:18 UTC 2000
  - Reenable/reimplement older login, but only use for ICQ UINs
    - This is a fairly ugly hack.  But...eh.  It works.
    - You'll need to remove the aim_sendconnack() before the
      call to aim_request_login.  It will now do it automatically
      if its needed.  (hint: ***CLIENT CHANGE***)

 - Wed Nov  8 02:23:25 UTC 2000
  - I'm trying out using a modified version of the kernel-doc scripts
     to help document the libfaim code a bit.  See aim_conn.c and 
     aim_tlv.c, as well as the scripts in utils/docs.

 - Wed Oct 25 17:52:20 UTC 2000
  - Cleanup warnings from newer gcc's (RH7.0)
    - Aparently newer gcc's don't let you do as strong of integer types
      on va_arg as older ones... So, let me know if this breaks on your
      compiler

 - Tue Oct 17 02:10:13 UTC 2000
  - Rearrange a few things

 - Mon Oct 16 20:56:59 UTC 2000
  - Remove some warnings

 - Mon Oct  9 04:24:56 CDT 2000 (jbm)
  - took care of the XXXs in aim_putuserinfo()
     - now use dynamic TLV count to send ("look ma, no magic!")
     - now send cap blocks
     - now send icq info if a numeric SN

 - Wed Oct  4 21:29:47 CDT 2000 (jbm)
  - Fix borked disconnect callbacks
  - Add aim_conn_in_sess()
  - A little housekeeping in faimtest ("fixing" order of printf, 
      remove a comment)