On Sun, Dec 28, 2003 at 09:09:42PM -0500, Jeff Trawick wrote: > it will be fine anyway :) it is okay to fail the test, as it just > brings a few extra instructions... it is only bad to pass the test > when in fact it should be failed > > some number of boxes will start failing the test now, and incur the > trivial amount of extra code in APR for resolving IPv4-mapped addresses
Ah how time changes things, the original patch I had was just #ifdef'd for DARWIN, a wholly simpler solution ;) "It does no harm on working systems and will fix the broken ones" - me 4 months ago. Curse whoever wrote that test (NB: may have been me :/). Does anyone still have a Darwin system that exhibits the broken behaviour? Does anyone know what getnameinfo() returns when passed ::127.0.0.1? If it's somethign useful like EAI_SYSTEM, EAI_FAIL or EAI_FAMILY that can be used to detect that particular failure mode (as opposed to just EAI_NONAME). -- Colm MacCárthaigh Public Key: [EMAIL PROTECTED]
