slidentd has been released in version 1.0.0
- there are quite a few bugfixes since 0.0.15, which i recommend to be removed
renaming the ebuild does the trick - perhaps the message in
pkg_postinst() could be cleaned up a bit:
something like this perhaps:
echo -e "\e[32;01m If you have supervise running, issue the following \033[0m"
echo -e "\e[32;01m command, to start your identd service:\033[0m"
echo ' # ln -s /var/lib/supervise/slidentd/ /service'
acutally, this needs changed...
AUTH_USERID=`id -u nobody`
AUTH_GRPID=`id -g wheel`
AUTH_USERID=`id -u nobody`
AUTH_GRPID=`id -g nobody`
it should also be emphasized, that slidentd prior to version 0.0.17 is not 100% RFC compatible, see CHANGES:
0.0.17 + honest_identd.8 manpage added
+ changes to use buffer.h and associated functions
+ eliminate lots of extraneous memory allocations by using buffer_put* where appropriate
+ ditch err_msg_prealloc et al (yippee!)
+ buggy newline append in slid_die (or was it slid_log?) removed and callers audited to add newlines
+ return \r\n as per RFC
0.0.18 + tcptable parser bug fixes
+ debug code debugged
0.0.19 + variable tcp table line lengths now correctly handled
+ uid >= 1000 bug fix
+ general tcptable parsing bug fixes
+ tcptable parsing changed to use buffer_get and friends and greatly simplified
1.0.0 + rfc removed
+ version incremented to 1.0.0
so upgrading is recommended.
other than that - slidentd-1.0.0 is running fine here :-)
Thilo, since you're the one who originally checked in the slidentd package, I'll go through with the changes you recommend here -- but I'll be relying on you if that's ok, to continue to maintain this (through me).
Please let me know if that's acceptable.
happy new year to you, seemant!
fine with me - i am usually reporting stuff that i use myself
OTOH - if i am the only one using this stuff i could keep it private and you'd just remove it altogether...
the group change is a bug introduced "back then" - it was supposed to add grsecurity support (being able to read from /proc needs wheel grouprights or something like that). but it didn't work as expected...
I've removed daemonutils as a runtime depend - doco seems to suggest that any inetd will do. As such virutal/inetd is a RDEPEND.
How does this work "ln -s /var/lib/supervise/slidentd/ /service"? I'm not sure its the right thing to do in FHS.
Any other problems let me know. I like seemant are trusting this one to you.
the "ln -s /var/lib/supervise/slidentd/ /service" part is for the daemontools startup in /service - it is definitivly in violation to the FHS...
its just there so people know that there is a daemontools startup script in
my change in comment #1 was a try to emphasize, that this is only applicable, when you have daemontools running...
Please either depend on virtual/inetd or on ucspi-tcp, not on both. The latter one includes tcpserver, so it should provide virtual/inetd (see Bug #62032). The virtual/inetd depend is redundant (as soon as the mentioned bug is fixed).
If slidentd strictly needs ucspi-tcp, just depend on that. If it works with any inetd-like service, depend only on virtual/inetd.
Thanks Sasha. - Fixed