Chris Lear wrote:
It seems that Jo wants autodetection to: 1) comply with the documentation 2) just work for most people 3) be easily fixable in other cases
Yes.
This, it seems to me, is exactly what it does.
Show me it working properly on a out-of-the-box rpm/ports config on a direct connect, no NAT system. (ie "most people")
> OK, maybe it doesn't work
in Jo Rhett's system. But defining "most people" as "people who do things like Jo Rhett" is suspect at best.
Actually, I'm using it bone stock from FreeBSD ports, so yeah - in this case my config is "normal" FWIW..
I can see a case for saying "autodetection can't possibly work in all cases, so disable it by default". But saying "it doesn't do what I expect, so it's broken" seems to show a disregard for the documentation and all the (lots of) historical discussions about the subject. Anyone who has seen spam hitting ALL_TRUSTED (as I have) can sort the problem out with reference to good documentation in minutes. That's if they are like me and installed spamassassin without really knowing anything about it. If they were more switched on, they would have read the documentation first.
Really? Then why is it the most common FAQ on the list (see archives) Why does it break for so very many people with normal configs? Why does it completely fail when installed in an old (pre-1990) class-A network? Eh?
Nobody cares what I expect. When the docs say "it should work for you" then I think it really should work for a majority of people. It's working for some weird subset of people that I can't quite figure out from reading the code.
-- Jo Rhett Network/Software Engineer Net Consonance
