[15:49:31]  <alindebe> dcantrell: 506384
[15:50:16]  <dcantrell^ hello
[15:50:30]  <alindebe> hello :D
[15:52:02]  <dcantrell^ 506384 is as-expected.  the owner of the setup
package asked during F-11 development that anaconda not write to
/etc/hosts.  basically, it's not supposed to be our job anymore.  let me
find that bug number
[15:52:51]  <alindebe> excellent
[15:53:20]  <dcantrell^ 491808 was the original
[15:55:15]  <dcantrell> it's not a dupe, but I wouldn't close it as
notabug.  it needs to go to another package, perhaps setup
[15:57:07]  <alindebe> ok
[15:57:11]  <alindebe> I'll send it to them, then
[16:00:09]  <dcantrell^ refer to the original bug too.  how to best handle
/etc/hosts is a topic that came up in a meeting I was in last week in the
office, so this is really more of an ongoing discussion
[16:00:21]  <dcantrell> initscripts gets involved, NetworkManager,
anaconda, setup, and so on
[16:00:41]  <alindebe> Yep, said we no longer write to /etc/hosts due to
other bug, sent on
[16:13:59]  <halfline> dcantrell: but it can't be setup's job since setup
just has a static file
[16:14:29]  <halfline> and it can't be networkmanager's job since you need
a resolvable internal hostname potentially before networkmanager is
started...
[16:15:24]  <halfline> hmm, well maybe that last part isn't true
[16:39:17]  ***  alindebe_ (n=alindebe@nat/redhat/x-107bf4397723f236) has
joined chat #anaconda.
[16:39:30]  ***  alindebe has left chat #anaconda (Nick collision from
services.).
[16:39:34]  ***  alindebe_ is now known as alindebe.
[16:51:20]  <dcantrell> halfline: yeah, it's probably not setup that
should own it.  unsure about NM owning it or not.  this was a discussion
we had last week, but no solution came up.  anaconda is writing out
HOSTNAME to /etc/sysconfig/network for now, but not touching /etc/hosts
[16:52:37]  <halfline> dcantrell: probably makes sense for anaconda to
write out /etc/hosts after setup is installed
[16:53:06]  * alindebe sulks
[16:53:17]  <alindebe> coulda had this discussion *before* I started
playing ping pong :P
[16:55:33]  <dcantrell^ sorry
[16:56:30]  <dcantrell> halfline: for some reason I thought we were doing
that at some point (earlier than what happened in 491808) and bugs were
opened to change it.  could be wrong
[16:57:17]  <halfline> funky
[16:57:45]  <halfline> maybe code just got rewritten/refactored so the
order changed (wrt to setup getting installed) unintentionally
[16:57:51]  <dcantrell> nonetheless, it's entirely reasonable.
[16:57:55]  <dcantrell> that's also entirely possible
[16:58:13]  <dcantrell> alindebe: assign it to me and put it back to
anaconda, por favor.  I want to look at this more once I get through my
endless to do list
[16:58:30]  <alindebe> ok
[16:59:25]  <dcantrell> halfline: I think I'm getting confused with the
bugs that you've opened that go down the path of "what do you want to call
the computer?" question rather than just hostname.
[17:00:25]  <halfline> different side of the same issue i guess
[17:00:50]  <halfline> fundamentally, for the computer to function it
needs a "hostname" it can resolve locally
[17:01:06]  <halfline> it doesn't necessarily need a name others can
resolve
[17:01:26]  <halfline> and the internal hostname doesn't need to make the
external hostname
[17:01:39]  <halfline> s/make/match/
[17:02:06]  <dcantrell> halfline: right.  adding the hostname entered by
the user to the localhost line has caused people to go ape in the past
[17:03:14]  <dcantrell> I'll take a look at it later though.  so many
things to process right now
[17:03:27]  <halfline> probably because there's some app out there that
does a lookup on the hostname to figure out which interface to bind to
[17:03:50]  <halfline> which is a totally broken thing to do
[17:04:40]  <dcantrell> yeah, I seem to remember those reports too.  ah
well
[17:38:08]  ***  denise has left chat #anaconda (Remote closed the
connection).
[18:09:29]  ***  rudi (n=rlandman@nat/redhat/x-fc459d79e89f0914) has
joined chat #anaconda.
[19:00:13]  <jeremy> dcantrell: we switched to writing it out before when
we switched to writing out all netconfig before.  which we need for things
like root-on-iscsi
[19:02:21]  <jeremy> at least that's what my fuzzy memory is saying.  if
you want me to add a comment to that effect to the bug in question I can
[19:52:58]  <dcantrell^ back now.  sure, feel free to add anything you
remember.  I do remember the hostname stuff changing around a good bit
because of rewriting the network code and/or other things needing it
changed