I'm going out on a limb stating that as a "fact" in the summary. I do not know of any reason that env-update should be called, and I can't find anyone else who has a good reason. There should not be any unprocessed env.d files as a result of the stage3 generation, so why do we tell users to run env-update? Is it just tradition or superstition? I bring this up because after the recent handbook change to use emerge-webrsync in place of the manual snapshot fetch/unpack, new users now get an ugly warning: !!! Invalid PORTDIR_OVERLAY (not a dir): '/usr/portage' !!! main-repo not set in DEFAULT and PORTDIR is empty. >>> Regenerating /etc/ld.so.cache... We are getting a couple of questions a day on this in #gentoo, I think we should either rethink the necessity of env-update, or move some steps around in the handbook so that env-update happens after emerge-webrsync. Thanks!
I just ran into this too... And I've already started a topic on this on the forum: http://forums.gentoo.org/viewtopic-p-7213912.html#7213912 ...
There doesn't seem to be a need for env-update. We need env-update when a /etc/env.d file is changed, as it will combine all those files into /etc/profile.env (which is sourced when you source /etc/profile). I did keep the env-update && source /etc/profile in the ppc bootloader section as perhaps the installations done prior to it create files there. However, for the regular installations (i.e. early in the process) there should not be a need for this. I also verified if a /etc/profile and /etc/profile.env file is provided by the stage3 files (checked for amd64, I assume it will be similar for the other architectures), which is the case, so we should be fine to just source /etc/profile.
The corrected paragraph talks about three steps, but only two are described.
Created attachment 334740 [details, diff] Patch which describes the third step
Thanks, applied
