[pptp-server] v0.8.7 released

[Matthew Ramsay]

Hiya all,

PoPToP v0.8.7 has been released...
You can download it here:
http://www.moretonbay.com/vpn/download_pptp.html

Important notes: It seems NT client support is broken... I'm not sure
exactly what version broke this. If anyone can prove me wrong though
please do tell (my testing is done with 98).

To address some issues Seth brought up:
v0.8.7 does not yet add much in the way of localip configuration support
(sorry Seth.. it's on its way soon). David Luyer submitted a patch with
many bug fixes... but also added support for PPPD to be responsible for
the local/remote ip definitions.. this may help.

Kevin has added more support to clean up stray CTRL connections.

Solaris/Slirp port has been patched in as well... although i may have
broken that a bit too (Harald?).
OpenBSD port is probably still broken (Peter?)

The complete ChangeLog is below. 

v0.8.6 -> v0.8.7
1st June, 1999

- GRE seperated from pptpctrl to support vforking
- adds link status detection (ie if a link goes down, we can figure it
out
    and deal with it) (Kevin)
- Solaris/Slirp port (Harald Vogt)
- cleaned up comments a lot (move towards C style to permit compiling in
    older compilers/increase portablity)
- standardized #ifndef #define #endif defines in header files
- stop inststr from nuking environment (hopefully)
- use longer argv[0] in exec()s to make inststr much nicer
- make inststr wipe args other than argv[0]
- #define to remove some debugging (PPTPD_DEBUG) and to remove the
    IP address allocation code so PPPD can be used to allocate IP
addresses
    (INTERNAL_IP_ALLOCATION)
- in pptpctrl.c, main()'s addrlen was uninitialized - yuck, was causing
    random variable overwriting
- clean up some wasteful memory copying and so on, as well as remove
some
    copies into small static buffers
- clean up some blank lines - increasing the amount of code visible one
    screen is good if it can be done without making the formatting ugly.
- use exit() not _exit() in pptpd - the fear of this closing fork()d
    filedescriptors is wrong.  both have the same file descriptor
closing
    properties.
- remove a potential leak of 2 filedescriptors in option parsing (checks
    of optional file names).
- miscellaneous EMBED support for syslog etc.
- many other misc changes.


That is all for now.

Cheers,
Matt.