Re: [PATCH 1/2] buildchroot: Do not call 'apt-get update'

[Alexander Smirnov <asmirnov@ilbers.de>]

Hi,

On 02/05/2018 08:25 PM, Jan Kiszka wrote:
> On 2018-02-05 18:06, Alexander Smirnov wrote:
>> On 02/05/2018 07:51 PM, Jan Kiszka wrote:
>>> On 2018-02-05 17:43, Alexander Smirnov wrote:
>>>> We should not do this to avoid local Packages.gz update.
>>>>
>>>> Signed-off-by: Alexander Smirnov <asmirnov@ilbers.de>
>>>> ---
>>>>    meta/recipes-devtools/buildchroot/files/configscript.sh | 1 -
>>>>    1 file changed, 1 deletion(-)
>>>>
>>>> diff --git a/meta/recipes-devtools/buildchroot/files/configscript.sh
>>>> b/meta/recipes-devtools/buildchroot/files/configscript.sh
>>>> index 9813c9a..458c94b 100644
>>>> --- a/meta/recipes-devtools/buildchroot/files/configscript.sh
>>>> +++ b/meta/recipes-devtools/buildchroot/files/configscript.sh
>>>> @@ -44,5 +44,4 @@ mount -t devtmpfs -o mode=0755,nosuid devtmpfs /dev
>>>>      #configuring packages
>>>>    dpkg --configure -a
>>>> -apt-get update
>>>>    umount /dev
>>>>
>>>
>>> Can you also explain why we (no longer?) need this? Just to ensure that
>>> there are no subtle issues waiting to be surfaced.
>>
>> Multistrap calls 'apt-get install' for the list of packages, so at this
>> point the '/var/lib/apt' folder is created. So the upstream state used
>> to generate buildchroot is cached.
>>
>> The line I removed in this patch, is called *after* rootfs generation,
>> so upstream state in cache could be possibly updated, do we need that?
> 
> As we are heading for a single update run, no. I just wanted to ensure
> that it's clear why we do not need it - and ideally also why we once
> thought we need it.
> 
> Jan
> 

I checked the patch which added thi and it was dated by 2016 and 
contains big blobs for various scripts. I guess it was merge from some 
existing example. I've double checked the multistrap Wiki page at 
debian.org and manual, there is no recommendation to have it in 
configscript. So due to it works without it - I'd like to drop it.

Alex