[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#599879: live-helper: Live Builder does not create the user account



On 12/10/10 05:45 AM, Jose R R wrote:
> On Tue, Oct 12, 2010 at 1:28 AM, Daniel Baumann <daniel@debian.org> wrote:
>> On 10/12/2010 10:15 AM, Jose R R wrote:
>>> I started with the same version for both and on the second and
>>> subsequent attempts building the image, the regular user was not
>>> created even if verified in the relevant config/* files.
>>
>> you can't 'verify' the user creation throuhg grepping config/*, again,
>> the user is created by live-config at runtime, not by live-build at
>> buildtime.
> 
> Faced with subsequent build attempts and *no* user created by the
> Debian Live scripts, I *explicitily* specified an user name of my own
> that I could grep in config/* files. The result was the same: no
> regular user name and no explicit name either.

You're still missing something here.  Yes, having no user created at
runtime is a problem (which Daniel adequately explained already).  But
no, your grep won't find any files associated with it!  That's because
the user is created when the system is booted, not when live-build runs.

>> i still claim its because of this:
>> http://live.debian.net/gitweb?p=live-build.git;a=commitdiff;h=f02a69126b802af2b3fdf79b851283a6d40e7c5a
>>
>> you were using live-build 2.0.1, but your target system did not get
>> live-config 2.0.7.
> 
> Whatever the reason, *that* procedure got rid of the regular user
> issue and now I have no problems with that aspect.

If you try random things you will get random results.

>>
>>> The only problem now is that the resulting image is
>>> not hardware agnostic. But again, this is an issue that reveals the
>>> lack of maturity of Debian Live builder.
>>
>> i don't understand, could you plase elaborate?
> 
> If an end user follows rudimentary instructions, the Debian Live
> scripts should provide satisfactory results. As it is now, one must
> dig into the workings of the Debian Live scripts to attempt to figure
> out where the issues are being introduced.

How was that an answer to the question?  What do you mean by "not
hardware agnostic"?

I agree that it is not always easy to tell at what layer a problem is
occurring, but that's because there are necessarily multiple layers, and
in this particular case, two: what happens at build time (handled by
live-build) and what happens at run time (handled by live-boot and
live-config). But if you're asking that the whole system always be
working at all phases of development ... well, I think you need to
readjust your expectations a bit.

>>
>>> I'm using scripts in x86
>>> hardware and not in x64 as Juan is doing.
>>
>> i386 or amd64 makes no difference here.
> 
> Of course, it should not -from a very specific and theoretic
> development environment. However in general hardware platforms there
> are issues unresolved/unfinished affecting the outcome.

No, quite concretely, i386 or amd64 makes no difference with regards to
the 'no user' issue caused by a *temporary* mismatch between live-build
on the build system and live-config in the image.  This has to do with
the migration process (which is made even more of a problem by the
freeze during release) which causes a delay for new versions entering
Squeeze and nothing whatsoever to do with the "maturity" of Debian Live.

Ben


Reply to: