[OE-core] [PATCH] tzdata: We shouldn't override the localtime if it is valid

Otavio Salvador otavio at ossystems.com.br
Fri Feb 8 12:36:39 UTC 2013

On Fri, Feb 8, 2013 at 10:27 AM, Tomas Frydrych
<tf+lists.yocto at r-finger.com> wrote:
> On 08/02/13 12:07, Otavio Salvador wrote:
>> On Fri, Feb 8, 2013 at 10:00 AM, Tomas Frydrych
>> <tf+lists.yocto at r-finger.com> wrote:
>>> On 08/02/13 11:55, Burton, Ross wrote:
>>>> On 8 February 2013 11:38, Otavio Salvador <otavio at ossystems.com.br> wrote:
>>>>>> Can you fix the code to symlink instead of copy?  systemd and more
>>>>>> needs a symlink so that it can identify the name of the timezone from
>>>>>> the link target.
>>>>> You mean inside the if?
>>>> I mean in general /etc/localtime should be a symlink and not a copy.
>>> /etc/localtime is a real file so that /etc is not containing a link to a
>>> file in /usr; this has historically been the case unless I am not
>>> mistaken. systemd should grow up and read the contents :).
>> In fact a lot of software expects it to be a symlink so it shouldn't be a copy.
> Lot of software does thing without understanding what it is doing,
> that's the nature of software. :) My point is that the code in the
> current recipe is not doing the unconditional copy by mistake, it is
> doing so intentionally so that local clock can work without /usr. Only
> systems that mandate the presence of /usr can use a symlink -- so this
> is not a bug fix but a change in an established system behaviour. If
> nothing else, the commit message should be clear in this regard.

I agree about the commit log. I'd for ignore the possibility to use
/etc without /usr as the applications running at very beggning will
start using GMT and move to timezone when /usr is ready. I see no
problem with that.

Ross, we are at same page? If yes I will send a v3 with improved commit log.

Otavio Salvador                             O.S. Systems
E-mail: otavio at ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854              http://projetos.ossystems.com.br

More information about the Openembedded-core mailing list