Man, I wonder if you break nvidia-kernel ebuilds on purpose. :-) Have you ever tested those ebuilds? The current stable one, nvidia-kernel-1.0.4496-r3.ebuild, references the function "is_kernel" which is not defined anywhere in the ebuild. It is however present in the 5536-r2 ebuild. This problem was mentioned in a previous bug I can't find in the open ones, so I think it was closed (without testing the ebuild, huh?) If calling an undefined function isn't enough, it is called with the wrong parameters! In fact it's called like "is_kernel 2.5" or "is_kernel 2.6", whereas the right way to call it (assuming the right way is the way the is_kernel() function works in the 5536-r2 ebuild) is "is_kernel 2 5" or "is_kernel 2 6". I hope I did not sound too harsh, but... ebuilds should be tested before being committed (and mm-sources-2.6.5 should be set as BLOCKER on nvidia-kernel due to Andrew's 4K stack patch...). :-) Reproducible: Always Steps to Reproduce: 1. emerge nvidia-kernel 2. 3.
OK technically this is my fault for not looking harder earlier when the 4363 problem was fixed. However Im pleased to say that as of 1500 UTC April 30 things have been committed back into working state. Closing bug.
