RPSL automation

Nick Hilliard nick at inex.ie
Wed May 20 10:56:22 UTC 2009


On 15/05/2009 18:50, Kevin Oberman wrote:
> Actually, there was a time when I used aoe quite a bit. It's been broken
> for so ling that I have long since learned to live without it. I would
> not mind seeing it back, but I don't see investing the time that would be
> required to get it working again. (Of course, if there are existing
> patches to it, then I'd like to see it stay.)

There's nothing really wrong with the idea of aoe/roe; it's just that their 
implementation is godawful.

I've created a new ticket on the irrtoolset trac with a patch which makes 
aoe compile (http://irrtoolset.isc.org/ticket/26).  It appears to run to 
some limited extent, although I haven't tested it in any meaningful way. 
If there are no objections to the patch, I would recommend committing it 
and then retiring aoe, roe and all of the extra baggage that is required 
for them to actually compile.

The reason for this suggestion is as follows.

- roe is completely broken and hasn't been updated to cope with the 
internal codebase change from Prefix* to MPPrefix*.  I'm not going to port 
roe to using mpprefixes, because it's not worth the effort.  So, roe should 
be removed unless anyone else is prepared to do something with it.

- aoe is still ipv4 only and does not support ipv6 or multicast. 
Furthermore, it generates rpsl which causes librpsl to barf.  I'm sure 
there are piles of other bugs in it too, because my patch is quick-n-dirty. 
  But more importantly, aoe is a horrible mishmash of ancient tcl, tk and 
badly structured c++ which make up an ugly, user-unfriendly, unmaintainable 
and buggy program.  I don't believe that it's worth supporting, and if 
there is some effort in future to create a system with equivalent 
functionality, then it would be better done as a javascript enabled webapp, 
or else using some other portable, window-capable scripting language (maybe 
java, ugh).  The purpose of committing a compilation patch is twofold: to 
provide the few people who really want to actually use the program with a 
recent version which at least compiles, and secondly to provide anyone who 
might want to maintain this program with a starting point.

The patch doesn't fix up the Makefile includes and libs required for aoe.

Nick



More information about the irrtoolset mailing list