updated spec files for release

[netatalk.git] / ChangeLog

($Id: ChangeLog,v 1.18 2001-03-07 20:39:24 rufustfirefly Exp $)

2001-03-07 15:30 EST  jeff b  <jeff@univrel.pr.uconn.edu>  

        Released 1.5pre5

        * distrib/rpm/netatalk-redhat.spec, distrib/rpm/netatalk-mandrake.spec:
        updated for 1.5pre5 release

2001-03-07 10:34 EST  jeff b  <jeff@univrel.pr.uconn.edu>  

        * etc/afpd/volume.c: changed VOLOPT_MAX to be 9 if FORCE_UIDGID is not
        defined (thanks to Axel Bringenberg <A.Bringenberg@srz-berlin.de>)

2001-03-07 10:14 EST  jeff b  <jeff@univrel.pr.uconn.edu>  

        * ChangeLog: started using timestamps in ChangeLog

        * etc/uams/uams_krb4/Makefile.am: fixed reference to send_to_kdc.c typo
        (thanks to Sebastian Rittau)

2001-03-06 13:40  Lance Levsen <l.levsen@printwest.com>

        * FAQ, README, README.ASUN, INSTALL.txt: Moved FAQ, AND READMEs to
        docs/, where they should be.

        * INSTALL/INSTALL.txt: Added ./INSTALL/INSTALL.txt

        * INSTALL/FAQ, README, README.ASUN: Moved README.ASUN, README,
        FAQ to ./INSTALL

2001-03-06 11:47  Andrew Morgan <morgan@orst.edu>

        * TODO: A few updates to papd entry.

        * README.MORGAN: Removed README.MORGAN because that information in
        now in papd's man page.

        * man/man8/papd.8.tmpl: Updated papd man page to match current
        code.  Added descriptions of authenticated printing and other new
        papd options.

2001-02-28 15:43  Marc J. Millar <itlm019@mailbox.ucdavis.edu>

        * libatalk/adouble/ad_open.c: AppleDouble directory creation
        debugging
        
2001-02-28  jeff b  <jeff@univrel.pr.uconn.edu>  

        * etc/afpd/directory.c, etc/afpd/file.c, etc/afpd/filedir.c,
        etc/afpd/unix.c, etc/afpd/unix.h, etc/afpd/volume.h,
        etc/afpd/volume.c, man/man5/AppleVolumes.default.5.tmpl: added
        "dropbox" to available option if DROPKLUDGE is used during
        compile 

2001-02-27  jeff b  <jeff@univrel.pr.uconn.edu>  

        * README: updated 1.5+ install instructions to include list of
        required and recommended packages

        * etc/uams/uams_*.c: cleanups, addition of CVS Id tag to C source

        * configure.in, acconfig.h: change USE_AFS to AFS to be the same as
        all of the defines in the codebase

        * etc/uams/uams_dhx_pam.c: fixed DHX login using this module (last
        patch made with syslog()'s didn't include any brackets) 
        (Bug #233756)

        * distrib/initscripts/.cvsignore: removed pulling of atalk

        * configure.in, etc/uams/Makefile.am: conditional compilation support
        for PGP UAM module using --enable-pgp-uam

        * configure.in, etc/uams/Makefile.am, etc/uams/uams_krb4/Makefile.am,
        etc/uams/uams_krb4/.cvsignore, etc/uams/uams_krb4/*.c: modifications
        for future Kerberos module reintegration

2001-02-26  jeff b  <jeff@univrel.pr.uconn.edu>  

        * configure.in: added /usr/local/ssl to list of SSL paths to check, to
        help kludge compilation on Mac OS X from Marcel <lammerse@xs4all.nl>

        * distrib/initscripts/rc.atalk.redhat.tmpl: adjusted to echo warning
        instead of dumping out if appletalk module not present, from
        Steven Karen <karelsf@users.sourceforge.net> (Bug #404087)

        * configure.in, contrib/timelord/timelord.c: applied patch from Wes
        Hardaker <hardaker@users.sourceforge.net> (Patch #402245), with
        suitable configure.in fixes

2001-02-23  jeff b  <jeff@univrel.pr.uconn.edu>  

        * etc/afpd/desktop.c, etc/afpd/codepage.c, etc/afpd/nls/makecode.c:
        patch from Axel Barnitzek <barney@users.sourceforge.net> to fix
        broken codepage support.

        * ChangeLog: started updaing ChangeLog with important patch/fix
        information, as it is *never* up to date.

        * configure.in, acconfig.h: implemented AFS configuration option
        patch from Wes Hardaker <hardaker@users.sourceforge.net>

        * VERSION: bumped up version to 1.5pre5, since 1.5pre4 was kind of
        paperbag-ish

        * autogen.sh: make libtoolize copy instead of linking files to
        avoid problems, thanks to Wes Hardaker <hardaker@users.sourceforge.net>

2001-02-20  jeff b  <jeff@univrel.pr.uconn.edu>

        Released 1.5pre4
        - Debian packaging in tree
        - Numerous Makefile/build fixes
        - .cvsignore implemented
        - Solaris build fixes

2001-01-02  jeff b  <jeff@univrel.pr.uconn.edu>

        * etc/afpd/uid.c, etc/afpd/uid.h, ...: added support for forcing
        uid/gid per volume for afpd

2000-09-22  Roland Schulz <rdschulz@abarrach.franken.de>

        * etc/afpd/volume.c (setvoltime): fix for multiple clients
        writing to same volume.

2000-02-28  a sun  <asun@asun.cobalt.com>

        * etc/afpd/directory.h (CNID_INODE): xor the inode a little
        differently. 

2000-02-23  a sun  <asun@asun.cobalt.com>

        * etc/afpd/volume.c (creatvol): / is a special case. you can't
        share it unless you give it a name.

2000-02-21  a sun  <asun@asun.cobalt.com>

        * distrib/initscripts/rc.atalk.redhat/cobalt: added changes to
        make redhat 6.x happier.

2000-02-17  a sun  <asun@asun.cobalt.com>

        * libatalk/adouble/ad_lock.c (adf_unlock): off-by-one error with
        lock removal. this + the log right below fix ragtime.

2000-02-16  a sun  <asun@asun.cobalt.com>

        * etc/afpd/fork.c (afp_bytelock): only error on bytelocks
        positioned at 0x7FFFFFFF if there's no resource fork.

2000-02-14  a sun  <asun@asun.cobalt.com>

        * libatalk/adouble/ad_lock.c: re-wrote locking bits so that
        allocations happen in blocks. added missing case that omnis
        database triggers.

2000-02-07  a sun  <asun@asun.cobalt.com>

        * bin/nbp/Makefile (install): make nbprgstr/nbpunrgstr with 700
        permissions. 

        * include/atalk/adouble.h (sendfile): change to deal with
        <sys/sendfile.h> 

2000-01-25  a sun  <asun@asun.cobalt.com>

        * etc/afpd/ofork.c: keep track of oforks being used for each
        directory so that we can update them if the directory tree gets
        modified. 

        * etc/afpd/directory.c (deletecurdir): remove dangling symlinks on
        delete. 

2000-01-24  a sun  <asun@asun.cobalt.com>

        * etc/afpd/directory.h (CNID): moved cnid assignment here along
        with helpful macros.

        * etc/afpd/directory.c: changed directory search to use red-black
        trees to improve balance. parent-child tree changed to circular
        doubly-linked list to speed up insert/remove times.  there's still
        one obstacle to actually freeing red-black tree entries. i need to
        add an ofork list to struct dir to minimize search times.

2000-01-18  a sun  <asun@asun.cobalt.com>

        * etc/afpd/directory.c (dirinsert): detect attempts to add
        pre-existing entries as just symbolic links. 

        * etc/afpd/filedir.h (CNID): moved inode-cnid assignment here and
        extended to directories.

2000-01-03  a sun  <asun@asun.cobalt.com>

        * etc/uams/uams_pam.c (PAM_conv): surround PAM_BINARY_PROMPT with
        an #ifdef.

        * etc/afpd/status.c (status_init): fixed a bunch of problems here
        that manifested under solaris 7.

        * etc/afpd/main.c (main): use FD_SETSIZE instead of FD_SETSIZE +
        1.
        

1999-12-27  a sun  <asun@asun.cobalt.com>

        * libatalk/util/getiface.c: moved interface detection code to here
        so that i can use if_nameindex() or getifconf() depending upon
        what's available.

1999-12-13  a sun  <asun@asun.cobalt.com>

        * libatalk/dsi/dsi_tcp.c (dsi_tcp_init): added if_nameindex()
        based interface code.

        * etc/afpd/afp_options.c (afp_options_parseline): added
        -server_quantum as an option. using hex would be a good idea.

        * libatalk/dsi/dsi_opensess.c (dsi_opensession): added bits to set
        the server quantum. by default, the server quantum is limited to
        1MB due to a bug in the os 9 appleshare client.

        * distrib/initscripts/rc.atalk.{cobalt,redhat}: surround nbp stuff
        with double quotes.

        * etc/uams/uams_dhx_pam.c (pam_changepw): added dhx-based password
        changing for pam.

1999-12-06  a sun  <asun@asun.cobalt.com>

        * etc/afpd/directory.c (setdirparams): don't error if we can't set
        the desktop owner/permisssions.

1999-11-04  a sun  <asun@asun.cobaltnet.com>

        * etc/afpd/fork.c (afp_openfork): had the ordering wrong on an
        openfork. 

1999-11-02  a sun  <asun@asun.cobaltnet.com>

        * etc/afpd/afp_dsi.c (afp_over_dsi): flush data for unknown dsi
        commands. 

1999-10-28  a sun  <asun@asun.cobaltnet.com>

        * etc/uams/*.c: return FPError_PARAM if the user is unknown. 

1999-10-27  a sun  <asun@asun.cobaltnet.com>

        * etc/afpd/fork.c (afp_read): if sendfile isn't supported, use the
        old looping method.

1999-10-25  a sun  <asun@asun.cobaltnet.com>

        * libatalk/nbp/nbp_unrgstr.c (nbp_unrgstr): fix nbp unregisters.

1999-10-21  a sun  <asun@asun.cobaltnet.com>

        * etc/afpd/Makefile (install): moved install of afpd earlier per
        suggestion by steven michaud.

1999-10-05  a sun  <asun@asun.cobaltnet.com>

        * etc/uams/uams_randnum.c (afppasswd): for ~/.passwd's, turn
        ourselves into the user so that nfs is happy.

1999-09-19  a sun  <asun@adrian5>

        * libatalk/netddp/netddp_open.c, nbp/*.c: only use the bcast stuff
        if it's on an os x server machine.

1999-09-15  a sun  <asun@adrian5>

        * libatalk/nbp/nbp_unrgstr.c,nbp_lkup.c,nbp_rgstr.c: os x server
        wants ATADDR_BCAST. that probably means that i need to do
        multihoming appletalk a little differently. bleah.

1999-09-09    <asun@asun.cobaltnet.com>

        * etc/afpd/directory.c (getdirparams), libatalk/adouble/ad_open.c
        (ad_open): mondo lameness. i forgot that directory lookups can be
        done with "." as the directory name. that was auto-hiding
        them. bleah. i also figured out which bit was the invisible bit
        for finderinfo information.

1999-09-06  Adrian Sun  <asun@glazed.cobaltnet.com>

        * etc/afpd/desktop.c (mtoupath): fixed a bug in codepage support
        that accidentally crept in.

1999-08-31  Adrian Sun  <asun@glazed.cobaltnet.com>

        * etc/afpd/quota.c (getfsquota): use group quotas in quota
        calculations if the user belongs to a single group. just use the
        user quotas if the user belongs to multiple groups.

        * etc/afpd/volume.c (getvolspace): added an options:limitsize to
        restrict the available space to 2GB. this is for macs running
        older versions of the operating system with newer versions of the
        appleshare client. weird huh?

        * etc/afpd/quota.c (uquota_getvolspace): bleah. 64-bit shifts
        don't get promoted in the same way as arithmetic operations. added
        some more casts to deal with that issue.

1999-08-24  Adrian Sun  <asun@glazed.cobaltnet.com>

        * man/man?/Makefile: don't re-build .tmp files if they already
        exist. this gets the install phase to work correctly.

1999-08-13  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/directory.c, file.c, filedir.c: illegal characters get
        AFPERR_PARAM. also, reject names with /'s in them if the nohex
        option is used.

1999-08-12  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/filedir.c,file.c,directory.c: changed error for
        illegal filenames to AFPERR_EXIST.

1999-08-11  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/desktop.h (validupath): if usedots is set, .Apple* and
        .Parent files are no longer valid file names.

        * etc/afpd/volume.c (volset): added usedots and nohex as
        options. usedots stops :hex translation of . files while nohex
        stops :hex translation of everything but . files. in addition,
        . files created on the unix side are by default hidden.

        * libatalk/adouble/ad_open.c: initialize more bits.

1999-08-10  a sun  <asun@hecate.darksunrising.blah>

        * distrib/initscripts/rc.atalk.redhat (WORKSTATION): use the
        actual name for nbp registration rather than ATALK_NAME.

        * sys/solaris/Makefile (kernel): make sure osdefs and machinedefs
        get used when building the kernel module.

        * sys/solaris: changed strings.h to string.h

1999-08-08  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (readvolfile): changed volume options into an
        array of structs to ease maintenance.

1999-08-05  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/status.c (status_init): change the default icon
        depending upon whether or not it's an ASIP or an AppleTalk
        connection. 

1999-08-04  Adrian Sun  <asun@glazed.cobaltnet.com>

        * etc/atalkd/main.c (setaddr): made a failure with setaddr a 
        little more informative.

1999-08-03  Adrian Sun  <asun@glazed.cobaltnet.com>

        * yippee. someone figured what was happening with the installation
        of the man pages. i got rid of a duplicate entry.

1999-08-02  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (readvolfile): added a per-file way of setting
        default options. it keys in on a :DEFAULT: label.

1999-07-30  a sun  <asun@hecate.darksunrising.blah>

        * moved rc.atalk.* scripts to distrib/initscripts.

1999-07-27  a sun  <asun@hecate.darksunrising.blah>

        * contrib/printing: added patch from job@uchicago.edu

        * etc/afpd/file.c: forgot to initialize struct ad in
        some places.

        * etc/afpd/nls/makecode.c: added an empty mapping.

        * etc/psf/Makefile (install): well cp -d didn't work either. just
        use tar.

1999-07-26  a sun  <asun@hecate.darksunrising.blah>

        * sys/solaris/tpi.c (tpi_attach): changed DDI_NT_NET to DDI_PSEUDO
        (from denny@geekworld.com).

        * distrib/rpm/netatalk-asun.spec (Summary): incorporated new spec
        and patch files from inoue.

        * sys/linux/Makefile (install-sysv): fixed up a bit.

        * etc/psf/Makefile (install): use cp -d instead of cp -a to make
        *bsd happier.

        * etc/afpd/afp_options.c (afp_options_parseline): reversed meaning
        of -icon. now it means to use the yucky bitmap instead of the
        apple icon.

        * bin/afppasswd/Makefile (all): add -Iinclude/openssl for
        afppasswd as well. 

1999-07-18  a sun  <asun@hecate.darksunrising.blah>

        * create links/mangle files in the compile rather than the install
        phase so that rpm will be happier.

1999-07-17  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/file.c (afp_createfile), directory (afp_createdir),
          filedir.c (afp_rename, afp_moveandrename): don't allow the
          creation/renaming of names with certain characters if mswindows
          compatibility is enabled.

1999-07-16  a sun  <asun@hecate.darksunrising.blah>

        * rc.atalk.redhat: incorporated chkconfig from inoue.

1999-07-15  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd/config.c (getifconf): wrap check against
        IFF_MULTICAST behind an #ifdef IFF_MULTICAST.

        * sys/netbsd/Makefile (LDSHAREDFLAGS): key in on machine type.

1999-07-11  a sun  <asun@hecate.darksunrising.blah>

        * contrib/ICDumpSuffixMap: added internet config perl script from
        inoue. 

        * contrib/printing: added contributed solaris printing scripts
        from job@uchicago.edu.

1999-07-10  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd/interface.h, rtmp.h: prototyped functions.

        * etc/atalkd/zip.c: converted bcopy's to memcpy's.

        * etc/atalkd/nbp.c,rtmp.c: added checks for the interface for
        dontroute cases.

        * etc/atalkd/main.c: converted bzero/bcopy to memset/memcpy.

1999-07-08  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/nbp/nbp_rgstr.c (nbp_rgstr): return EADDRINUSE if the
        address already exists.

1999-07-06  a sun  <asun@hecate.darksunrising.blah>

        * rc.atalk.redhat: changed netatalk.config to netatalk.conf

1999-07-05  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd/nbp.c (nbp_packet): add interface to nbp struct. this
        is so that we can filter by interface in the future. however, it
        doesn't seem to work that well right now. bleah. 

        * etc/atalkd/main.c: fixed up dontroute option so that it doesn't
        screw up atalkd.conf. also, we need to do a bootaddr if dontroute
        is set. 

        * libatalk/atp,nbp,netddp; bin/aecho,nbp,getzones,pap;
        etc/papd,afpd: accept -A <ddp address> as an option so that you
        can specify the address to be used on a multihomed server. for
        papd, you use the 'pa' option in papd.conf.

1999-07-04  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd/config.c (parseline): initialize parseline properly
        so that we don't get extraneous junk.

        * etc/afpd/afp_options.c (afp_options_parseline): do
        gethostbyaddr/gethostbyname's for -ipaddr and -fqdn.

        * etc/atalkd/config.c (getifconf/readconf): check to see if the
        supported device can support appletalk addresses. either continue
        or exit depending upon whether or not it's auto-configed.

1999-07-03  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/afp_options.c (afp_options_parse): -I (-[no]icon) will
        toggle the volume icon so that it uses the apple icon instead.

        * etc/afpd/config.c (AFPConfigInit): added more logic for the
        -proxy option. here are the rules: -proxy will always try to
        create a DDP server instance. by default, the proxy server will
        still allow you to login with an appletalk connection. to prevent
        that, just set the uamlist to an empty string.

1999-07-02  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/status.c (status_netaddress): added support for fqdn
        (not available in the appleshare client yet).

1999-07-01  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/config.c (DSIConfigInit): application code for proxy
        setup. it's the -proxy option.

        * libatalk/dsi/dsi_init/tcp.c (dsi_init/dsi_tcp_init): added
        support for proxy setup.

1999-06-30  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/filedir.c (afp_rename): fixed up some error
        codes. quark express should be happier.

        * etc/afpd/uam.c (uam_afpserver_option): added
        UAM_OPTION_HOSTNAME. use this to set PAM_RHOST. i just got a
        report that setting that fixes pam on solaris machines.

1999-06-28  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/ofork.c (of_alloc): report out of forks in syslog..

        * etc/afpd/enumerate.c (afp_enumerate): close an opendir leak. 

        * include/atalk/{dsi,asp}.h: make cmdlen and datalen ints.

        * etc/afpd/fork.c (afp_write): fixed up error condition.

1999-06-26  a sun  <asun@hecate.darksunrising.blah>

        * etc/uams/Makefile (install): changed install location of uams.

        * sys/linux/Makefile (install-sysv): always install redhat
        script. netatalk.config script only gets installed if it's not
        there already.

1999-06-23  a sun  <asun@hecate.darksunrising.blah>

        * rc.atalk.redhat: merged in redhat contrib rpm rc.atalk script.

        * etc/afpd/afp_options.c (afp_options_init): changed default
        maxusers to 20.

1999-06-22  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/config.c (DSIConfigInit): truncate options->server to
        just the server name here.

        * etc/afpd/volume.c (volxlate): made $s return something
        meaningful no matter what.

        * libatalk/adouble/ad_sendfile.c (ad_readfile): freebsd sendfile
        wants an off_t.

1999-06-20  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (volxlate): added variable substitution. if it
        doesn't understand the variable, it just spits it back out.
        
        (creatvol): display truncated volume name if it's too long.

        * sys/{generic,solaris}/Makefile: added NO_CRYPTLIB option to deal
        with oses that have -lcrypt but shouldn't use it.

1999-06-11  a sun  <asun@hecate.darksunrising.blah>

        * include/atalk/afp.h: added comments to FPErrors.

        * etc/afpd/enumerate.c (afp_enumerate): make FPEnumerate do some
        more error checking.

        * include/atalk/util.h: server_lock() returns pid_t. 

1999-06-10  a sun  <asun@hecate.darksunrising.blah>

        * README.ASUN: added location for both ssleay and openssl.

        * etc/uams: moved install to LIBDIR/uams. "uams_*" now means "uam
        server." in the future, there will be "uamc_*." changed the shared
        library names to match.

        * include/atalk/atp.h,nbp.h: forgot to include <sys/cdefs.h>

        * etc/uams/Makefile: openssl-0.9.3c uses <openssl/*.h> so add that
        to the include path.

        * sys/{solaris,ultrix}/Makefile: just use -I../sys/generic instead
        of doing a link.

        * include/atalk/uam.h, etc/uams/uam_*.c, etc/afpd/uam.c: added uam
        type field. do type check.

        * etc/uams/uam_*pam.c: added a couple more error codes. 

1999-06-08  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/nls/Makefile (codepage.h): make sure that a link to
        codepage.h gets made.

        * libatalk/*/Makefile: make sure that the profiled directory gets
        created.
        
        * etc/afpd/directory.c (afp_mapname): removed an extraneous line
        that was causing mapname to fail.

1999-06-07  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd/main.c (main): added a note to check the syslog if
        atalkd can't be setup.

        * sys/linux/Makefile: added -DNEED_QUOTACTL_WRAPPER to the list of
        auto-detected #defines.

1999-06-06  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c (afp_write): argh. i moved things around a
        little too much and ended up with an uninitialized eid. strangely,
        the compiler didn't complain. simplified bits a little as
        well. also, FPWrite was returning the wrong error messages. on
        64-bit filesystems, the offset can wraparound. so, report a disk
        full error if that's going to happen. egcs-19990602 gets one
        memcpy right and another wrong on my udb. bleah.
        
        (afp_read): fixed the error messages here as well.

1999-06-05  a sun  <asun@hecate.darksunrising.blah>

        * Makefile, sys/generic, sys/{ultrix,solaris}/Makefile: create
        some links on the fly if they're missing.

        * etc/afpd/directory.c (copydir): fixed a leaking opendir and
        re-arranged a little.

1999-06-04  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd: prototyped everything here and moved the FP functions
        into include files.

        * libatalk/util/bprint.c: moved all of the bprints to here.

        * libatalk/asp, include/atalk/asp.h: prototyped asp functions.

        * include/atalk/atp.h, libatalk/atp: prototyped atp functions. 

        * libatalk/nbp, include/atalk/nbp.h: added prototypes for nbp
        functions. 

        * bin/afppasswd/Makefile (afppasswd): fixed a misspelling in the
        install phase.

        * bin/afppasswd/afppasswd.c: added -a option so that root can add
        new users. turned all of the options into bits. added newlines to
        each entry.

1999-06-03  a sun  <asun@hecate.darksunrising.blah>

        * sys/freebsd/Makefile: turn on sendfile support if running on a
        FreeBSD 3+ machine.

1999-06-02  a sun  <asun@hecate.darksunrising.blah>

        * etc/uams/uam_dhx_pam.c: fixed memory freeing part of pam
        conversation function.

        * sys/*/Makefile: check at make time to see if -lrpcsvc and
        -lcrypt should be included in the appropriate places. 

1999-05-28  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/file.c (deletefile): added more error checking here as
        well. 

        * etc/afpd/directory.c (renamedir): added a couple a few more
        error bits. 

        * sys/sunos/Makefile: sunos should really work now. 

1999-05-27  a sun  <asun@hecate.darksunrising.blah>

        * include/atalk/afp.h: added in a couple new error codes (one
        deals with password setting policy, the other with maximum logins
        by any user).

        * etc/afpd/fork.c (afp_openfork): try to re-open files on
        read-only volumes as read-only.

1999-05-26  a sun  <asun@hecate.darksunrising.blah>

        * sys/solaris/Makefile: fixed a few bobbles here. solaris uses
        uname -p. other oses seem to use uname -m for the same information.

        * etc/uams/uam_pam.c (pam_changepw): added check for same
        password. 

        * etc/uams/uam_randnum.c (randnum_changepw): added in cracklib and
        same password checks.

        * sys/osx/Makefile: moved the os x server stuff into its own build
        directory. 

        * sys/linux/Makefile, sys/solaris/Makefile: key in on OSVERSION
        and MACHINETYPE for some stuff.  

1999-05-25  a sun  <asun@hecate.darksunrising.blah>

        * sys/sunos/Makefile: various bits to make stuff work with sunos
        again. 

1999-05-25  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/file.c (copyfile): only copy the header file if newname
        is specified. 

        * etc/afpd/directory.c (copydir): make sure to balk if the
        directory already exists. in addition, make sure to preserve the
        timestamps. 

1999-05-24  a sun  <asun@hecate.darksunrising.blah>

        * bin/afppasswd/afppasswd.c: global password updating utility for
        the randnum authentication method. 

1999-05-22  a sun  <asun@hecate.darksunrising.blah>

        * etc/uams/uam_randnum.c (afppasswd): added in global password
        file for the randnum authentication method. it looks for a .key
        file as well to handle encryption.

        * etc/afpd/afp_options.c (afp_options_parseline): added
        -passwdfile as an option so that you can specify a global randnum
        password file if desired.

        * etc/afpd/volume.c (readvolfile): we now have rwlist and rolist
        as an AppleVolumes.* option. if the user is in the rolist, the
        volume gets set as readonly. if there's a rwlist, and the user
        isn't in it, the volume also gets set as readonly.

1999-05-21  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_lock.c (ad_fcntl_lock): plug a leak if we
        can't allocate the reference counting variable.

        * etc/uams/uam_*.c: make sure that uam_setup returns an error
        code. 

1999-05-19  a sun  <asun@hecate.darksunrising.blah>

        * include/atalk/paths.h (_PATH_LOCKDIR): added os x server's
        /var/run as the lock file directory.

        * etc/afpd/fork.c (afp_write): kanehara@tpk.toppan.co.jp reported
        a problem with FPWrite getting a request count of 0. that's
        fixed. 

        * etc/afpd/Makefile: bleah. for some reason, pam doesn't like to
        load itself from a shared library. i've compensated by linking it
        into afpd again.

        * etc/uams/uam_dhx_passwd.c: okay. DHX now works. something's
        still screwy with the dhx_pam stuff though.

1999-05-18  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/uam.c (uam_getname): i forgot that getname modified the
        username to fit what's in pw->pw_name if necessary.

1999-05-16  a sun  <asun@hecate.darksunrising.blah>

        * etc/uams/uam_dhx_passwd/pam.c: almost ready versions of the DHX
        authentication method. i'm still missing a little info to get it
        all right.

        * bin/megatron/nad.c (nad_header_read): if there isn't a mac name,
        create it from the unix name. 

        * bin/megatron/megatron.c (megatron): oops. need to turn fdCreator
        and fdType into strings.

1999-05-16  a sun  <asun@pelvetia>

        * etc/afpd/uam.c (uam_afpserver_option): changed the interface a
        little. now, you pass in an int * if you want to either get/set
        the size of the option. added in UAM_OPTION_RANDNUM for generic 
        (4-byte granularity) random number generation.

        * etc/afpd/switch.c: added afp_logout to preauth_switch.

1999-05-15  a sun  <asun@hecate.darksunrising.blah>

        * bin/megatron/macbin.c (bin_open): make error message for
        macbinary files more informative.
        
        (test_header): added more macbinary tests. it now has a workaround
        for apple's incorrectly generated macbinary files.

1999-05-14  a sun  <asun@hecate.darksunrising.blah>

        * sys/solaris/Makefile: added shared library generation bits.

        * etc/uams: moved server-side uams here. 

        * include/netatalk/endian.h: fixed some solaris bits. 

        * etc/afpd/config.c (configfree): don't do an asp_close. instead,
        do an atp_close and free the asp object. oh yeah, as afpd needs
        to export symbols to its modules, make sure you don't do anything
        more exciting than strip --strip-debug with it.

1999-05-12  a sun  <asun@hecate.darksunrising.blah>

        * various places that use sigaction: zero out struct sigaction so
        that we don't send something confusing. also make sure that we
        don't set a timer unless we already have a sigaction set.

        * etc/afpd/fork.c (afp_openfork): don't error on trying to open an
        empty resource fork read-only. also, added back in the bit of code
        that prevented locks from being attempted on non-existent resource
        forks.

        * etc/afpd/afp_options.c (getoption): added a uamlist commandline
        option (-U list). 

        * libatalk/netddp/netddp_open.c: don't bind if nothing was passed
        in. 

        * libatalk/nbp/nbp_unrgstr.c (nbp_unrgstr): oops. forgot to
        convert this over to use by the netddp interface.

1999-05-12  a sun  <asun@pelvetia>

        * etc/afpd/uam.c: os x server's runtime library loader is
        braindead. as a result, i've switched to using an exported struct
        with the uam's name.

        * bin/aecho,getzones: changed these to use the netddp interface.

        * libatalk/nbp/nbp_rgstr.c,unrgstr.c: fixed more leaky bits.

        * libatalk/netddp: abstracted the ddp interface to netddp. besides
        the prior socket-driven interface, there's now an os x server
        interface. so, instead of calling socket/sendto/recvfrom, you call
        netddp_open/netddp_sendto/netddp_recvfrom.

1999-05-11  a sun  <asun@pelvetia>

        * libatalk/nbp/nbp_lkup.c: oh my. nbp_lookup was fd leaky if there
        was a problem.

        * etc/atalkd/main.c (main): make sure that if -dontroute is
        selected for all but one interface, that interface also gets
        -dontroute set.

1999-05-10  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/auth.c: re-wrote to deal with plug-in uams. it's much
        smaller than it used to be.

1999-05-09  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/uams/uam_guest.c, uam_pam.c, uam_passwd.c,
        uam_randnum.c: uam modules. these should probably be moved out of
        afpd (and into something like etc/uam_server) when the printing
        stuff gets uam support.

1999-05-08  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/uam.c: interface to user authentication modules.
        it should eventually be moved to libatalk, but that's not
        necessary until the printing uam stuff is done. everything is from
        the server-side perspective, but that's only because there aren't
        any client-side uses right now.

        * libatalk/util/module.c: generic interface to run-time library
        loading functions. right now, the dlfcn family and os x server's
        NS-style way of doing things are the ones understood. in addition,
        there's a DLSYM_PREPEND_UNDERSCORE for those systems that need it.

        * libatalk/asp/asp_write.c (asp_wrtcont): log both the read and
        write part of write continuations.

1999-05-07  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd: added the ability to turn off routing for particular
        interfaces. specify -dontroute for each interface that you don't
        want to route.

1999-05-06  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/auth.c: got rid of global clrtxtname and switched to
        using obj->username.

1999-05-04  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/dsi/dsi_write.c (dsi_write): dsi_write could loop
        forever if there's a problem while it's being used. that's fixed.

1999-05-01  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/directory.c (renamedir,copydir,deletedir): added bits
        so that renaming a directory works across filesystems.

1999-04-27  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c (getforkparams): report mtime if it's > than
        what's stored in the header file.

        * config/afpd.conf: incorporated a patch by Stefan Bethke to make
        afpd.conf more understandable.

        * sys/solaris/if.c: many of the firstnet/lastnet bits weren't
        endian converted. that's fixed.

        * libatalk/adouble/ad_lock.c (adf_find(x)lock): F_RD/WRLCK aren't
        necessarily ORable, so use ADLOCK_RD/WR instead.
        
        (ad_fcntl_unlock): erk. fixed a typo that had the resource fork
        unlock accidentally getting the data fork locks.

1999-04-24  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c (afp_openfork): always try to create a resource
        fork if asked.

1999-04-21  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_open.c, ad_read.c/ad_write.c, ad_flush.c:
        turned the mmapped header stuff into and #ifdef
        USE_MMAPPED_HEADERS option.

        * libatalk/adouble/ad_open.c (ad_header_read): darn. i forgot that
        the hfs fs doesn't currently have mmappable header files. rather
        than implement that, i just reverted back to a modified version
        of the old way of reading headers.

1999-04-15  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c (afp_bytelock): byte locks become read locks on
        read-only files.
        
        (afp_openfork): deal with read-only data forks that don't have
        corresponding .AppleDouble files. we can't really do anything with
        deny locks in this case. just make sure that read locks are set.

        * etc/afpd/file.c (getfilparams): oops. got the parentheses wrong
        around FILPBIT_FINFO. 

        * etc/afpd/fork.c (afp_read): as we share open files now, check
        for fork type against of_flags instead of just checking to see if
        the file is open. this fixes a bug that caused resource forks to
        get filled with data fork information.

1999-04-09  a sun  <asun@porifera.zoology.washington.edu>

        * sys/generic/Makefile: AFP/tcp now compiles on irix with quota
        support.

1999-04-09  a sun  <asun@mead1.u.washington.edu>

        * sys/generic/Makefile: AFP/tcp now compiles on aix with quota
        support.

1999-04-09  a sun  <asun@saul6.u.washington.edu>

        * sys/generic/Makefile: AFP/tcp part now compiles on digital unix
        with quota support enabled.

1999-04-08  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c, fork.c, file.c, directory.c, filedir.c,
        config/AppleVolumes.default: added read-only volume option.

        * etc/afpd/quota.c (uquota_getvolspace): modified for os x
        server. 

1999-04-03  a sun  <asun@hecate.darksunrising.blah>

        * bin/megatron/macbin.c (bin_write): only pad if we need to do so 
        (from <jk@espy.org>).   
        (bin_header_write/read): fixed up screwed up file date
        generation/reading with macbinary files. 

        * bin/megatron: changed all of the bcopy/bzero/bcmp's to 
        memcpy/memset/memcmp's. added macbinary III support.

        * bin/megatron/macbin.c (bin_open): added --stdout as an option so
        that we can stream macbinary file creation to stdout.

        * bin/megatron/megatron.c: incorporated information patch (--header
          and --macheader) from <fmorton@base2inc.com>.


1999-04-02  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd: whee! there are no more bcopy/bcmp's in this
        directory. 

        * libatalk: changed the bcopy/bzero's to memcpy/memset's. added in
        dummy ints for some of the files that can get compiled to empty
        objects. check for the type of msync() available as well.

1999-03-31  a sun  <asun@hecate.darksunrising.blah>

        * INSTALL/README.GENERIC: added information for a generic
        architecture. It includes the information needed to get netatalk
        to compile on a random unix platform.

        * etc/afpd/quota.c: moved the quota stuff here so that we can
        #ifdef it out on a machine without quota support.

1999-03-30  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_lock.c: reference count the locked ranges as
        well. this prevents multiple read locks on the same byte range
        from disappearing if one user disappears.
        
        (ad_fcntl_lock): here are the current rules for file
        synchronization:
           1) if there's a appledouble header, we use the beginning
              of that for both data and resource forks.
           2) if there isn't, we use the end of the data fork (or past the
              end on 64-bit machines)
        
1999-03-28  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_open.c (ad_refresh): okay. mmapping the
        appledouble entry parts is done. 

        * libatalk/cnid/cnid_add.c (cnid_add): prevent anyone from adding
        in an illegal cnid.

1999-03-27  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_open.c (ad_refresh): started making the
        appledouble header parsing more generic so that we can read in
        arbitrary appledouble header files. i just mmap the parts that we
        need. 

1999-03-22  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/file.c (afp_copyfile): return the correct error
        response on a failed copy. also, error if the file is already open
        to prevent problems with locks. we really need to ad_lock
        this during the copy

1999-03-21  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (readvolfile): switched volume options to
        using ':' as a delimiter as that's one of the characters that's
        not allowed as part of a mac filename.
        (volset): changed access to allow/deny

        * etc/afpd/auth.c (noauth_login): make sure that the username gets
        set. 

1999-03-17  a sun  <asun@hecate.darksunrising.blah>

        * NOTE to myself: jeremy allison said that samba uses refcounts to
        prevent close() from killing all the byte locks. so, i've started
        converting afpd to using refcounting as well. luckily, we already
        have of_findname, so we know when files are open. in cases where
        files are already open, this will replace an ad_open with a lookup
        into a hash table.

        * etc/afpd/directory.c (getdirparams/getfilparams): check for
        NULL names when getting directory/file names.

        * etc/afpd/directory.{c,h} (DIRDID_ROOT/DIRDID_ROOT_PARENT): make
        sure these are always in network byte order.

1999-03-15  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c (afp_openfork): okay, fixed the file
        synchronization bits. now, we use two bytes to do the read/write
        stuff. when access is needed, a read lock is attempted. if a deny
        lock is needed, a write lock is attempted. we even handle the
        access None mode now by saving the access modes.

        * etc/afpd/fork.h (AFPFORK_ACCMASK): started adding bits so that
        we can obey all of the file synchronization rules.

        * etc/afpd/fork.c (afp_bytelock): got the meaning of the clearbit
        reversed. with helios lantest's lock/unlock 4000 times test, it
        looks like i get <1 second overhead on my machine when using byte
        locks. NOTE: this will get a little worse when serialization gets
        added. in addition, 0x80000000 only works for 64-bit machines. i
        reserve the last allowable bit for 32-bit machines. 

        actually, both 64-bit machines and 32-bit machines use 0x7FFFFFFF
        now as i'm able to trigger a kernel oops in linux with the 64-bit
        code. 
        
        (afp_read/afp_write): make sure to use the same offset when doing
        a tmplock.

1999-03-14  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_lock.c: i went and implemented a bunch of
        stuff to get byte locks to work correctly (except for the
        serialization) only to discover that files that use byte locks
        also request a deny write mode. luckily, byte locks only cover up
        to 0x7FFFFFFF. so, i'll just use 0x80000000 for the
        synchronization locks.

1999-03-08  a sun  <asun@hecate.darksunrising.blah>

        * sys/{*bsd,ultrix,solaris,linux}/Makefile (depend): surround 
        DEPS with double quotes so that multiple defines work.

1999-03-06  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_write.c, ad_read.c: make off off_t in size. 

        * libatalk/adouble/ad_flush.c (adf_fcntl_relock), ad_lock.c
        (adf_fcntl_locked): okay. fcntl locks now check against multiple
        programs on the same machine opening the same file. current
        problems with the mechanism that i don't want to fix unless
        necessary: 
                1) there's a race during the relock phase. serialization
                   would solve that.
                2) it assumes that each fd only locks a single contiguous
                   range at a time. keeping a list of locked ranges would
                   solve that. 
        
        also, i changed the default to using fcntl locks. if the above two
        are really necessary, i'll probably switch to something a little
        more featureful like the berkeley db's lock manager.

        (note to myself: stuff new from asun2.1.3 from 1999-03-03)
        
1999-03-05  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_lock.c: got rid of the endflag checks to
        reduce system calls a little. 

        * etc/afpd/auth.c (getname): do a case-insensitive compare on the
        login name as well.

        * sys/solaris/Makefile: added 64-bit solaris patch from
        <jason@pattosoft.com.au>.

1999-03-03  a sun  <asun@hecate.darksunrising.blah>

        * include/netatalk/endian.h: make solaris 2.5 complain less.

        * bin/adv1tov2/adv1tov2.c, libatalk/adouble/ad_open.c (ad_v1tov2):
        fixed a couple problems with the adv1tov2 stuff.

1999-02-26  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (afp_openvol): erk. the volume password gets
        aligned along an even boundary.

1999-02-23  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (readvolfile): added volume password support. 

1999-02-14  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd/multicast.c (addmulti): added FreeBSD's data-link
        layer multicast setting bits.

        * libatalk/adouble/ad_open.c (ad_v1tov2): make sure to stick in
        prodos field info when converting.

        * rc.atalk.redhat: added pidof checking in case the machine
        crashes. also added rc.atalk.wrapper to the redhat rc script
        installation. 

1999-02-07  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c (afp_setforkparams): make sure to do better
        error detection here and more fully report error conditions.
        
        (flushfork): make sure to flush the header if necessary (rfork
        length changed or modification date needs to be set).

        (afp_write): ugh. this wasn't returning the right values for the
        last byte written if the endflag was set. in addition, it was
        setting the modification date. that should be left to FPCloseFork
        and FPFlush(Fork). this fixes a problem that shows up with
        QuarkXPress. 

        NOTE: as of now, changes to the rfork info are the only things
        that aren't flushed immediately when altered. 

        * etc/afpd/fork.c (get/setforkparams), ofork.c: what ugliness. we
        need to report bitmap errors if we try to fiddle with the wrong
        fork. i added an of_flags field to keep things sorted.

        * libatalk/adouble/ad_open.c (ad_v1tov2): oops. in all of the
        movement, i forgot to make sure that the pre-asun2.2.0 features
        still compile.

1999-02-06  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/filedir.c (afp_moveandrename): make sure to save the
        old name even when it's a directory.

        * globals.h: added oldtmp and newtmp to AFPObj to reduce the
        number of buffers used. use these when needed in afp_* calls.

        * etc/afpd/directory.c (deletecurdir): delete stray .AppleDouble
        files when deleting a directory.

1999-02-05  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/file.c (afp_createfile): fixed a hard create error
        check bug. 

        * fixed up a few bobbles in the netatalk-990130 merge.

        * the noadouble option should be pretty much implemented. here's
        how it goes:
                when a directory is created, the corresponding
                .AppleDouble directory is not.
                
                comments, dates, and other file attributes will get
                silently ignored and not result in the creation of a new
                .AppleDouble directory.

                however, if anything possessing a resource fork is copied
                into that directory, the corresponding .AppleDouble
                directory will be created. once this happens, any
                other file in the directory can acquire an AppleDouble
                header file in the future.
                
1999-02-03  a sun  <asun@hecate.darksunrising.blah>

        * merged in the rest of netatalk-990130.

        * sys/solaris: merged in netatalk-990130 changes.

        * etc/afpd/file.c,fork.c,desktop.c libatalk/adouble/ad_sendfile.c:
        tested and fixed the sendfile bits on linux. it looks like linux
        doesn't permit socket -> file sendfiles yet.

        * etc/afpd/fork.c (afp_read): we can't stream FPRead's with
        newline character checking on.

1999-02-02  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c (afp_flush), ofork.c (of_flush): FPFlush
        operates on a per-volume basis.

1999-01-31  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/file.c (copyfile): sendfile bits added here also.

        * etc/afpd/desktop.c (afp_geticon): added sendfile bits here as
        well. 

        * libatalk/adouble/ad_sendfile.c (ad_writefile/ad_readfile):
        implemented sendfile bits. currently, there's support for linux
        and freebsd. unfortunately, freebsd's implementation doesn't allow
        file->file or socket->file copies. bleah.

1999-01-30  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/file.c (setfilparams), directory (setdirparams),
        volume.c (volset): added in the beginnings of a NOADOUBLE option
        for those that don't want AppleDouble headers to be created by
        default. it doesn't really work that well right now.

1999-01-29  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_open.c (ad_v1tov2): separated v1tov2 bits
        from ad_refresh. made broken v1 header checking the default. fixed
        broken v1 date checking. now, it just checks to see if the v1
        MDATE is > than st_mtime by 5 years.

        * etc/afpd/directory.c: make date setting alter directory dates as
        well. 

1999-01-24  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/afp_dsi.c (alarm_handler,afp_over_dsi): added a
        CHILD_RUNNING flag to prevent afpd from timing out on long copies.

1999-01-21  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (afp_openvol), libatalk/cnid/cnid_nextid.c:
        shift the beginning of the fake did's if necessary.

        * libatalk/adouble/ad_open.c (ad_refresh): fixed screwed-up date
        detection code.

        * libatalk/cnid/cnid_add.c,cnid_open.c,cnid_close.c: made some
        changes so that the CNIDs will still work even when the .AppleDB
        directory is read-only. if you're still allowed to create files on
        these volumes, you'll get a temporary id for those.

1999-01-20  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/cnid/{cnid_close.c,cnid_open.c}: added bits so that log
        files get cleared out on cnid_close() if it's the last user for a
        volume.

1999-01-18  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (afp_setvolparams): added FPSetVolParms. this
        allows us to set the backup date on the server.

        * etc/afpd/file.c (afp_exchangefiles): whee! we now have
        FPExchangeFiles. we also have FPDeleteID, so that only leaves us
        with FPCatSearch to do.

1999-01-16  a sun  <asun@hecate.darksunrising.blah>

        * sys/solaris/ddp.c (ddp_rput): added a couple htons()'s for the
        net addresses.

1999-01-11  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (volset, afp_openvol): you can now specify a
        dbpath= in AppleVolumes.* for the CNID database.

        * libatalk/adouble/ad_open.c (ad_refresh): if we build in an
        appledouble v1 environment, turn on v1compat by default.

1999-01-10  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_open.c (ad_refresh): added v1compat option
        to handle broken ad headers.

        * etc/afpd/file.c (setfilparams): we need to make sure that we
        flush the file if we've created it even if there's an error.  the
        magic number/version don't get saved if we don't.

        * etc/afpd/appl.c, etc/afpd/directory.c, etc/afpd/desktop.c: added
        DIRBITS to mkdirs.

1998-12-30  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (afp_openvol): got rid of unnecessary v_did.

        * etc/afpd/file.c (afp_resolveid, afp_createid): added these two
        in. 

        * well, what do you know? the cnid stuff compiles in. 

1998-12-29  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c, directory.c, file.c, filedir.c, volume.c,
        enumerate.c: added in stubs for CNID database conditional on
        AD_VERSION > AD_VERSION1.

        * etc/afpd/nls/makecode.c: added iso8859-1 mapping.

1998-12-27  a sun  <asun@hecate.darksunrising.blah>

        * bin/adv1tov2/adv1tov2.c: turn non-printable ascii characters
        into hex code as well.

1998-12-21  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/auth.c: fixed FPChangePW for 2-way randnums.

1998-12-15  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/fork.c (read_file/write_file): do crlf translation in
        both directions so that we can recover from problems if
        necessary. 

1998-12-14  a sun  <asun@hecate.darksunrising.blah>

        * bin/adv1tov2/adv1tov2.c: small utility program that recursively
        descends a directory and converts everything it sees into
        AppleDouble v2.

1998-12-13  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_flush.c (ad_rebuild_header): moved the
        header rebuilding here so that ad_refresh can use it.

        * libatalk/adouble/ad_open.c (ad_refresh): added locking to v1->v2
        conversion.

        * bin/megatron/asingle.c: yuk. removed all of
        the duplicate stuff here and had it use the #defines in adouble.h.  

        * libatalk/adouble/ad_open.c (ad_refresh): finished v1 -> v2
        conversion routine. we still need a shortname creator and a cnid
        database for the v2 features to be useful. 

1998-12-11  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/adouble/ad_open.c (ad_refresh): punt if we get a file
        that we don't understand.

1998-12-10  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/desktop.c (utompath,mtoupath): simplified the codepage
        stuff. also made sure to lower/upper the 8-bit characters as
        well. 

        * libatalk/util/strdicasecmp.c: the casemapping had a few wrong
        characters. 

        * etc/afpd/fork.c (getforkparams): make sure that the ROpen/DOpen
        attribute bits are in the correct byte ordering.

1998-12-09  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (volset): made prodos an option to options=
        flag. also added crlf as an option.

        * libatalk/adouble/ad_open.c (ad_refresh): fix up times if
        necessary. 
        (ad_open): deal correctly with 0-length files by treating them as
        newly created. 

        * etc/afpd/volume.c (getvolparams), file.c (get/setfilparams),
        fork.c (getforkparams), directory.c (get/setdirparams): finished
        adding appledouble version 1 and 2 date conversion. also added
        attribute setting. 
        
        * etc/afpd/volume.c (getvolparams): make sure to flush the header
        file if we needed to fiddle with it. 

        * libatalk/adouble/ad_date.c, ad_attr.c: date/attribute
        setting/retrieval code. 

        * libatalk/adouble/ad_open.c (ad_open): initialize date
        structures here instead of elsewhere. 

1998-12-07  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/directory.c, fork.c, volume.c, file.c: added unix<->afp
        time conversion code. 

1998-12-05  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (volset): changed prodos setting to
        prodos=true. 

1998-12-04  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/volume.c (volset): okay. you now have the following
        options to casefold: lowercase,uppercase,xlatelower,xlateupper
                tolower    -> lowercases everything in both directions
                toupper    -> uppercases everything in both directions
                xlatelower -> client sees lowercase, server sees uppercase
                xlateupper -> client sees uppercase, server sees lowercase

        NOTE: you only should use this if you really need to do so. this
        and the codepage option can cause mass confusion if applied
        blindly to pre-existing directories.
        
1998-12-03  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/desktop.c (utompath,mtoupath), etc/afpd/volume.h: added
        multiple options to casefold. bits 0 and 1 deal with MTOU, and
        bits 2 and 3 deal with UTOM. i did it that way so that you can
        casefold in one direction only or in both directions if
        desired. needless to say, setting both bits for UTOM or MTOU
        doesn't make any sense. right now, toupper takes precedence in
        MTOU, and tolower takes precedence in UTOM.

1998-12-02  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/afp_options.c (afp_options_*): added -uampath and
        -codepagepath to the list of available options. they specify the
        directories to look for uam modules and codepages,
        respectively. currently, -uampath doesn't do anything.

        * etc/afpd/volume.c (readvolfile): spruced up options to
        AppleVolumes files. now you can have mtoufile=<codepage.x>,
        utomfile=<codepage.y>, casefold=<num> for volumes.

        * etc/afpd/desktop.c (utompath,mtoupath): added
        codepage/casefolding support. casefold is currently an int that
        could have multiple actions. right now, i just lowercase in
        mtoupath and uppercase in utompath.

        * etc/afpd/ofork.c (of_alloc, of_findname, of_rename): added vol
        as an additional specifier so that we don't have problems with
        the same path names on multiple volumes. 

1998-11-29  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/volume.c (getvolparams): added AFP2.1 volume attribute
        bits. 

1998-11-24  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/atalkd/config.c (readconf, getifconf): added IFF_SLAVE to
        prevent atalkd from trying to use a slave channel.

1998-11-23  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/volume.c (getvolparams): we shouldn't set the custom
        icon bit by default on the root directory. that screws up pre-OS 8
        systems.

1998-11-19  a sun  <asun@purgatorius.zoology.washington.edu>

        * libatalk/dsi/dsi_getsess.c (dsi_getsession): ignore SIGPIPE's
        so that we can gracefully shut down the server side.

        * etc/afpd/afp_dsi.c (afp_over_dsi), etc/afpd/afp_options.c,
        libatalk/dsi/dsi_getsess.c (dsi_getsession),
        libatalk/asp/asp_getsess.c (asp_getsession): made the tickle timer
        interval an option (-tickleval <sec>).
        
        * etc/afpd/afp_dsi.c (afp_dsi_timedown): added child.die so that
        we don't stomp on a shutdown timer. libatalk/dsi/dsi_read.c,
        dsi_write.c both save/restore the old timer, so they don't really
        care what the actual value is. 
        
1998-11-18  a sun  <asun@purgatorius.zoology.washington.edu>

        * due to the recent obsession with bug fixing and low-level dsi
        cleanups, i've decided that this should really be asun2.1.1
        instead of asun2.1.0a.

1998-11-17  a sun  <asun@purgatorius.zoology.washington.edu>

        * libatalk/dsi/dsi_tcp.c (dsi_tcp_open): moved the afpd connection
        announcement here from etc/afpd/afp_dsi.c. 
        
        * libatalk/dsi/dsi_stream.c: moved all of the read/write functions
        into here as they're pretty generic. now, the protocol specific
        stuff only handles open and close. 

        * etc/afpd/fork.c (afp_read/write), desktop.c (afp_geticon),
        file.c (copyfile), include/atalk/dsi.h (dsi_writefile/readfile):
        added initial stubs for sendfile support. i still need to go
        through and calculate the appropriate offsets to use. 

        * libatalk/dsi/dsi_read.c, dsi_write.c: disable the interval timer
        instead of just ignoring it.

        * etc/afpd/desktop.c (afp_geticon), etc/afpd/fork.c (afp_read),
        libatalk/dsi/dsi_read.c (dsi_readinit, dsi_readinit): modified the
        dsi_read interface to return errors so that i can kill things
        gracefully. 

1998-11-16  a sun  <asun@purgatorius.zoology.washington.edu>

        * libatalk/dsi/dsi_tcp.c (dsi_tcp_send/dsi_tcp_write): erk. read()
        and write() treat a return of 0 differently. 

1998-11-16  a sun  <asun@hecate.darksunrising.blah>

        * libatalk/dsi/dsi_read.c (dsi_readinit): make sure to stick in
        the error code.

1998-11-15  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/fork.c (afp_read): re-ordered some of the checks here
        to return earlier on 0-sized files.

1998-11-13  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/afp_dsi.c (afp_over_dsi): moved the dsi->noreply toggle
        check to here from dsi_cmdreply.

1998-11-11  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/atalkd/zip.c (zip_packet): make sure to copy multicast zone
        back out. (reported by Michael Zuelsdorff <micha@dolbyco.han.de>)

1998-11-09  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/directory.c (getdirparams): changed unknown bit reply
        code to AFPERR_BITMAP instead of AFPERR_PARAM.

1998-11-06  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/enumerate.c (afp_enumerate), directory.c (renamedir):
        fixed a couple of failed realloc leaks.

        * etc/afpd/filedir.c (afp_moveandrename, afp_rename): added bits
        to deal with case-insensitive, case-preserving filesystems.

1998-10-30  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/auth.c: fixed randnum password changing check. 

1998-10-27  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/atalkd/main.c: add a check for SIOCATALKDIFADDR if
        SIOCDIFADDR fails.
        
        * etc/afpd/volume.c (getvolparams): ad_open had the wrong
        parameters. 

        * etc/afpd/unix.c (setdeskowner): added a little extra space to
        prevent buffer overflows here.

1998-10-26  a sun  <asun@purgatorius.zoology.washington.edu>

        * sys/linux/Makefile: fixed PAM message.

        * sys/solaris/Makefile: make failure to ln -s a non-fatal error. 

        * etc/papd/session.c, bin/pap/pap.c: changed sequence count to
        wrap from 0 -> 1 instead of from 0xFFFF -> 1.

        * etc/afpd/filedir.c (afp_rename, afp_moveandrename): actually, we
        should check against the entire unix name.
        
1998-10-21  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/filedir.c (afp_rename, afp_moveandrename): make sure
        to check against mac name.

1998-10-19  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/auth.c (afp_changepw): make password changing deal
        correctly with "real" user names. also, moved seteuid() to before
        the pam_authenticate() bit as shadow passwords need that.

        * etc/afpd/enumerate.c (afp_enumerate): make sure to check the mac
        name against MACFILELEN.

1998-10-16  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/file.c (renamefile), filedir.c (afp_rename),
        directory.c (renamedir): use strndiacasecmp() for AFPERR_SAMEOBJ
        checks. also make sure test occurs before checking to see if the
        destination exists.

1998-10-15  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/auth.c (afp_changepw): fixed a bit of brain damage. i
        forgot that password changing needs root privileges to work.

        * etc/afpd/auth.c (PAM_conversation): the supplied code was
        incorrect. i cleaned it up a bit.

        * sys/linux/Makefile: fixed the installation bits.

1998-10-14  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/auth.c (afp_changepw): don't kill the connection here
        if there's a problem.

1998-10-10  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/enumerate.c, fork.c, ofork.c, file.c,
        globals.h, directory.c, auth.c: #defined MACFILELEN and used
        that. also made sure that files > MACFILELEN never show up.

1998-09-25  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/{afpd,papd,atalkd}/bprint.c (bprint): got rid of the
        spurious pointer dereference.

        * etc/afpd/ofork.c (of_alloc/of_rename): allocate the max mac file
        length so that we don't need to realloc.

        * etc/afpd/filedir.c (afp_rename, afp_moveandrename): make sure to
        return AFPERR_BUSY if the dest has an ofork open.

        * etc/afpd/file.c (renamefile), directory.c (renamedir), filedir.c
        (afp_rename): return AFPERR_SAMEOBJ if source == dest

1998-09-21  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd: went through and looked for places that needed to use
        curdir instead of dir. i think i have them all right now.

        * etc/afpd/filedir.c (afp_moveandrename): wasn't keeping track of
        curdir correctly. what this really means is that cname should be
        fixed to return everything it changes as opposed to changing a
        global variable. 

1998-09-19  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/config.c (configinit): do the right thing if
        AFPConfigInit fails.

1998-09-18  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/config.c (ASP/DSIConfigInit, configfree): how 
        embarrassing. i wasn't doing refcounts correctly.

1998-09-17  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/nfsquota.c (getnfsquota): ultrix uses dqb_bwarn instead
        of dqb_btimelimit.

        * sys/ultrix/Makefile: ultrix understands the old rquota format. 

        * etc/afpd/ofork.c (of_findname): erk. forgot to only search in
        the current directory. 
        (of_rename): erk. changed it to handle renaming a file that has
        been opened multiple times.

        * etc/atalkd: made sure that if you don't specify -router, things
        are as they were before. 

1998-09-13  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/status.c (status_flags): forgot to turn on password
        setting if randnum passwords are being used.

1998-09-11  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/unix.c (setdirmode): erk. make sure only to setgid on
        directories. 

        * bin/aecho/aecho.c (main): incorporated -c <num> (ala ping) patch
        from "Fred Lindberg" <lindberg@id.wustl.edu>.

1998-09-03  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/directory.c (afp_closedir, afp_opendir): added these in
        for more AFP 2.0 compliance. unfortunately, apple's appleshare
        client doesn't like non-fixed directory ids.

1998-08-31  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/volume.c (accessvol): the accessible volume list can
        now be controlled by groups and usernames. just use something of
        the following form: @group,name,name2,@group2,name3

        NOTE: you can't have any spaces, and the parser forces you to
        include all options. in this case, there are some apple II options
        that need to be entered. they need to go away soon anyway.

        * etc/afpd/auth.c (getname): oops. i forgot to copy the gecos
        field into a temporary buffer before calling strtok.

1998-08-29  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd/main.c (as_timer), rtmp.c (rtmp_delzones): when the 
        last router on an interface goes down, we need to delete the
        interface's zone table.

1998-08-28  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/status.c (afp_getsrvrinfo): although it's never used,
        i've added this in to increase AFP compliance.

        * etc/afpd/auth.c (afp_getuserinfo): added FPGetUserInfo -- this
        should make afpd compatible with OS 8.5's Nav Services.

        * etc/atalkd/config.c,main.c: -router now specifies router mode
        with any number of interfaces.

1998-08-27  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/atalkd/main.c (as_timer): well, i figured out how to set up
        atalkd as a single-interface router. now, you can get zones with
        only single interfaces! i'm only waiting on wes for the approved
        configuration toggle.

1998-08-26  a sun  <asun@purgatorius.zoology.washington.edu>

        * libatalk/adouble/ad_lock.c, include/atalk/adouble.h: turned the
        ADLOCK_* into real #defines and added translations in the
        lock-type specific functions. this should make it easier to
        recompile things without screwing up.

1998-08-26  a sun  <asun@hecate.darksunrising.blah>

        * etc/atalkd/nbp.c (nbp_packet): forgot to handle another local
        zone case.

1998-08-25  a sun  <asun@hecate.darksunrising.blah>

        * etc/afpd/status.c (status_server): changed status_server to
        use only the obj part of obj:type@zone-style names.

        * etc/atalkd/nbp.c (nbp_packet): unregistering wasn't handling
        zones properly. it was matching on anything but the actual zone.

1998-08-18  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/auth.c (clrtxt_login): added pam_set_time(PAM_TTY) so
        that solaris' pam session setup doesn't crap out.

1998-08-17  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/atalkd/multicast.c (zone_bcast): fixed to do the right thing
        with zip multicast info.

1998-08-15  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/nfsquota.c: made the old-style rquota fields dependent
        upon -DUSE_OLD_RQUOTA and defined that for sunos. also included
        <sys/time.h> for ultrix breakage.

1998-08-13  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/filedir.c (afp_rename), etc/afpd/ofork.c (of_rename): i
        knew that speeding up of_findname would be useful. in any case, i
        discovered the source of yet another small AFP non-compliance that
        was confusing WordPerfect. on an afp_rename, we also need to
        rename the corresponding ofork. i've added an of_rename() to do
        this.

1998-08-13  a sun  <asun@hecate>

        * etc/afpd/ofork.c (of_dealloc,of_alloc): sped up dealloc by
        sticking refnum in ofork.

1998-08-12  a sun  <asun@hecate>

        * etc/afpd/fork.c (afp_openfork): added already open attribute
        bits.

        * etc/afpd/ofork.c: added a hashed of_findname.

1998-08-06  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/fork.c (afp_openfork): fixed a problem with opening
        forks from read-only non-appledouble media.

1998-07-23  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/afs.c (afs_getvolspace), etc/afpd/volume.c
        (getvolspace): modified them to treak afs like the other
        getvolspaces w.r.t. VolSpace.

1998-07-21  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/unix.c (mountp): erk. i forgot that symlinks are often
        used for filesystems. nfs quotas sometimes failed as a
        result. that's fixed now.

1998-07-20  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/auth.c (login): added a -DRUN_AS_USER #define so that
        it's simple to run the server process as a non-root user.

1998-07-17  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/nfsquota.c (callaurpc, getnfsquota), volume.h: it turns
        out that i was opening lots of sockets with callaurpc. now, the
        socket gets saved and reused.

        NOTE: quota-1.55-9 from redhat 5.x doesn't return the correct size
        for rquota's bsize. unless fixed, rquota will report incorrect
        values. 
        
1998-07-08  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/uam/README: added some preliminary ideas on a
        plug-in uam architecture. in addition, this should allow arbitrary
        afp function replacement. eventually, auth.c should get a
        bit cleaner.

1998-07-07  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/nfsquota.c: added headers and redefined a couple
        structure fields so that sunos4 compiles.

        * libatalk/compat/rquota_xdr.c: compile if we're using glibc <
        2. this should get redhat 4.2 machines. NOTE: they're still
        missing librpcsvc.a, so they'll need to remove that from the
        etc/afpd/Makefile. 

1998-07-06  a sun  <asun@purgatorius.zoology.washington.edu>

        * libatalk/compat/rquota_xdr.c: it turns out that solaris is
        missing a couple functions needed for rquota support. here they
        are. 

        * etc/afpd/unix.c (mountp): fixed the nfs detection for
        solaris. we still need bsd and ultrix.

1998-07-05  a sun  <asun@hecate>

        * include/atalk/adouble.h: marked out space for appledouble v2. 

1998-07-04  a sun  <asun@hecate>

        * etc/afpd: plugged up some ad_open leaks. made sure that we don't
        get negative numbers for length fields and such.
        
1998-07-04  a sun  <asun@hecate>

        * etc/afpd/nfsquota.c (getnfsquota): added nfs rquota
        support. Robert J. Marinchick <rjm8m@majink1.itc.virginia.edu>
        provided the initial bits from the bsd quota program. 

        * etc/afpd/unix.c (getquota): made getquota call getfsquota or
        getnfsquota depending upon the type of fs. 

        * etc/afpd/unix.c (mountp/special): munged mountp and
        special to return either the nfs mount point or the fs
        device. set the vol->v_nfs flag if it's nfs.

        * etc/afpd/volume.c (getvolspace): xbfree and xbtotal will now
        honor results returned from uquota_getvolspace.
        
1998-06-29  a sun  <asun@purgatorius.zoology.washington.edu>

        * etc/afpd/file.c (copyfile): mmapping the file copy only helps on
        multiple copies. as that's not the case here, i've reverted to
        just doing read + write.
        
1998-06-28  a sun  <asun@purgatorius.zoology.washington.edu>

        * sys/linux/Makefile: fixed the redhat-style atalk
        installation. also, it doesn't over-write an existing
        /etc/rc.d/init.d/atalk file.

        * etc/afpd, libatalk/adouble/ad_write.c: removed <sys/file.h> and
        </usr/ucbinclude/sys/file.h> so that we rely upon adouble.h.

1998-06-19  a sun  <asun@purgatorius.zoology.washington.edu>

        * changed sys/linux/Makefile to install a redhat-style sysv atalk
        script instead of the bsd one.

        * include/atalk/adouble.h: moved same-name list stub to struct
        ad_adf. 
        
Thu Jun 18 18:20:28 1998  a sun  <asun@purgatorius>

        * changed to asunx.y.z notation as i was getting tired of
        increasing numbers. as this version is undergoing a fairly
        substantial overhaul, i bumped it to 2.1.0. don't ask why asun1.0
        never existed. i just started at 2.0.

        * ofork (etc/afpd/{ofork.c,ofork.h,fork.c}: put in skeleton code
        for hashed-by-name oforks and oforks which group by name to help
        with byte locks and of_findname.

        * adouble (include/atalk/adouble.h): started implementing
        appledouble v2. mostly, i added bits to headers. v2 has wonderful
        bits that should make prodos support much less clunky, allow
        persistent dids, and allow msdos support.

        * finder info: added bits to directory.c and file.c describing the
        actual contents of finder info (from IM: Toolbox
        Essentials). also, set default directory view to an allowed value
        thanks to a suggestion from the appledouble v2 specs. that should
        help with read-only media.

        * etc/afpd/{directory.c,volume.c,afs.c,directory.h}: added
        DIRDID_ROOT and DIRDID_ROOT_PARENT so people know that these did's
        are reserved.
        
Wed Jun 17 11:54:49 1998  a sun  <asun@purgatorius>

        * well, i'm starting a changelog as i keep forgetting what i've
        done. 

        * locks: revamped them to be more in line with what should
        happen. currently, i've reverted back to flock-style locks as i'll
        need to stick in more code to get fcntl-style locks to work
        properly. specifically, i think modifying of_alloc/of_dealloc to
        keep track of locks, fds, and names should solve the problem with
        fcntl locks being process-specific and not fd specific.