RE: [NeoStats-Devel] [Commits] r159 - branches/3.0/libopm

Justin Hammond wrote:
> > Justin Hammond wrote:
> > > Well, considering it was broken before as well... :)
> > 
> > As mentioned on list and the forum on more than one 
> occasion, I have 
> > been running 3.0 in production for some time so that I have 
> a running 
> > real life test on each update. Thanks to libevent I have 
> not been able 
> > to update this since it would remove my ability to run 
> OPSB. OPSB was 
> > broken due to the premature introduction of libevent which is a 
> > different issue to the current status where even configure fails. :(
> > 
> 
> If you look at the changes made, I've removed libopm from 
> OPSB. 

I noticed that but since I did not know whether there was still some
requirement given that configure failed, I could not fix it myself. Since
there were no futher updates, I assumed you had maybe missed a commit or
something rather than you had intentionally made a partial commit hence my
concern.

> Previously libopm handled a lot of the socket 
> code/read/write etc etc and with the introduction of libevent 
> and the socket update, was very much duplicating what we have 
> now introduced. This really only struct me when I was working 
> through the libopm code trying to figure out the best way to 
> have it support libevent... I realised that we can do it with 
> a lot less work and duplication.

OK. Had you not noticed, I had and would have removed a lot of the libopm
stuff sooner had it not been one of those items awaiting socket/libevent
updates. One I can cross of my list :)

> What I commited tonight is not complete. Its 90% there. I 
> have to write the actual proxy protocol support and read 
> functions, but its not a huge amount of work. The remaining 
> work is around just cleaning up after scans, and should be 
> trivial to do.

OK cool. 

> You might also notice that I removed the dns blacklist 
> support from OPSB.
> This is partly due to the fact that people want to have 
> multiple DNS blacklist support, so intend to write a separate 
> module for this as well...
> (since it will no longer just be tied to Blitzed OPM list, 
> its not necessary proxy specific anymore, some might want to 
> use the TOR or spam lists as
> well)

The separate blacklist module is definitely an idea we agree on and as you
say should resolve a lot of feature requests for multiple blacklists. There
is also a mail from DNB in the archives with a list he suggested for
SecureServ type checks and I guess we can use it to work out the SecureServ
RBL framework you suggested recently.

> > > My homedrive is giving me smart errors, so I'm scared to leave 
> > > anything on it at the moment that isn't backed up. This 
> is work in 
> > > progress.
> > 
> > I know it is work in progress but based on your emails I 
> was under the 
> > impression you were fixing up socket stuff including opsb and 
> > implementing the config stuff last weekend. Is any of this going to 
> > happen any time soon?
> > 
> 
> The customers that pay the bills for the servers that host 
> the NeoStats
> infra come first... Sorry, but if I don't keep them happy, 
> then I got bigger
> problems than just a hard drive failure. I'll be putting in a 
> couple of
> hours each day this week on each (opsb and config)

There are always priorities in life but it is a simple courtesy to let
others know what is going on. I just wish you had.

Hopefully you will get on ok with things now and some good news, OPSB now
compiles for Win32. I don't know whether it runs or not, I just ran a
compile test on my main system then tested the Win32 one and fixed up the
include dirs and it linked :)

Mark.