public inbox for openembedded-core@lists.openembedded.org
 help / color / mirror / Atom feed
From: Robert Yang <liezhi.yang@windriver.com>
To: Mike Looijmans <mike.looijmans@topic.nl>,
	Richard Purdie <richard.purdie@linuxfoundation.org>,
	<openembedded-core@lists.openembedded.org>
Subject: Re: [PATCH 1/8] bitbake.conf: set CONFFILES by default
Date: Mon, 18 Jan 2016 10:21:48 +0800	[thread overview]
Message-ID: <569C4C3C.1080606@windriver.com> (raw)
In-Reply-To: <569A248F.9030500@topic.nl>



On 01/16/2016 07:07 PM, Mike Looijmans wrote:
> On 16-01-16 06:36, Robert Yang wrote:
>>
>>
>> On 01/15/2016 07:12 PM, Richard Purdie wrote:
>>> On Thu, 2016-01-14 at 18:29 -0800, Robert Yang wrote:
>>>> There are a lot of binary packages have files in /etc/, but only a
>>>> few
>>>> of them have set CONFFILES, more than 180 packages are not set in a
>>>> world build.  So treat all the files in /etc/ as CONFFILES as Debian
>>>> does:
>>>> - All the files in /etc/ are CONFFILES
>>>> - Move the file out of /etc/ if it is not a conffile
>>>> - If /etc/foo can't be moved out, and is changed during
>>>>    runtime, then create a symlink /etc/foo -> /var/foo
>>>>
>>>> [YOCTO #8436]
>>>>
>>>> Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
>>>> ---
>>>>   meta/conf/bitbake.conf |    1 +
>>>>   1 file changed, 1 insertion(+)
>>>
>>> We have discussed this before and we decided that setting CONFFILES
>>> globally seemed like a bad idea?
>
> That is how I recall it. Please please please don't do it. It will wreak havoc
> on all kinds of systems.
>
>> Sorry, but I can't find any emails which said this was a bad idea.
>> If we don't set CONFFILES by default, maybe we need mark a lot of
>> CONFFILES inside the recipes. I think that mark the CONFFILES under
>> /etc/ doesn't hurt anything, but help a lot when doing upgrade on
>> target:
>>
>> 1) When the old file is not changed, it will be replaced by new file.
>> 2) When the old file is changed, it will ask the user what to do (dpkg)
>>     or save a bak (rpm).
>
> In a perfect world, yeah.
>
> In the world other people have to live in, the GUI has to shut down to free up
> resources to get the upgrade to complete successfully. This means you cannot ask
> the user sitting on the couch holding the remote staring at a static "please
> wait upgrade in progress" screen any questions.
>
> Then if you could ask him questions, he'll just have no clue as to what to answer.
>
> You can create "backup" files all you like, but 99% of the users will never ever
> see the file system, and do not know how to rename, edit or copy a file. Which
> on the other hand does not stop them from trying, and they're quite used to the
> situation that a simple reinstall will bring things back in working order if
> they somehow "got it wrong".
>
> In the world I live in, things happen that change files that shouldn't be
> changed. Power outage (could be just the cleaning lady pulling the plug) just to
> name one.
>
> Manually fixing things up is fine if you're an experienced developer or power
> user. But remember that the majority of the end users are less tech savvy, and
> just use a "box" with software that happens to be built by OpenEmbedded, which
> they never heard of.


For apt-get, you can configure dpkg to use non-interactive when upgrade or
dist-upgrade:
--force-confnew or --force-confold

For ipkg, I think that you can use: --force-maintainer

For rpm, save a bak is the default action.

// Robert

>
>>> What has changed since the last discussion?
>
> Nothing relevant...
>
>


  reply	other threads:[~2016-01-18  2:21 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-01-15  2:29 [PATCH 0/8] bitbake.conf: set CONFFILES by default Robert Yang
2016-01-15  2:29 ` [PATCH 1/8] " Robert Yang
2016-01-15  6:27   ` Mike Looijmans
2016-01-15  6:37     ` Robert Yang
2016-01-16 10:51       ` Mike Looijmans
2016-01-18  1:33         ` Robert Yang
2016-01-15 11:12   ` Richard Purdie
2016-01-16  5:36     ` Robert Yang
2016-01-16 11:07       ` Mike Looijmans
2016-01-18  2:21         ` Robert Yang [this message]
2016-01-15  2:29 ` [PATCH 2/8] recipes-bsp: cleanup CONFFILES Robert Yang
2016-01-15  2:29 ` [PATCH 3/8] recipes-connectivity: " Robert Yang
2016-01-15  2:29 ` [PATCH 4/8] recipes-extended: " Robert Yang
2016-01-15  2:29 ` [PATCH 5/8] recipes-core: " Robert Yang
2016-01-15  2:29 ` [PATCH 6/8] recipes-devtools: " Robert Yang
2016-01-15  2:29 ` [PATCH 7/8] recipes-graphics: " Robert Yang
2016-01-15  2:29 ` [PATCH 8/8] recipes-multimedia/sato/support: " Robert Yang
  -- strict thread matches above, loose matches on Subject: below --
2015-10-09  2:44 [YOCTO 2.1] [PATCH 0/8] set CONFFILES to /etc/* by default as Debian does Robert Yang
2015-10-09  2:44 ` [PATCH 1/8] bitbake.conf: set CONFFILES by default Robert Yang
2015-10-09 14:19   ` Martin Jansa
2015-10-09 14:56     ` Robert Yang

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=569C4C3C.1080606@windriver.com \
    --to=liezhi.yang@windriver.com \
    --cc=mike.looijmans@topic.nl \
    --cc=openembedded-core@lists.openembedded.org \
    --cc=richard.purdie@linuxfoundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox